Every model

Most models are exactly what the label says. The 126 where the measurement adds something the label leaves out, or leaves a question open, come first.
- Models measured
- 337
- Distinct fingerprints
- 69
- Worth a look
- 126
measurement had something to addwithdrawnconfirmed the labelvariant or alias
Showing all 337 entries.
AionLabs: Aion 3.5aion-labs/aion-3.5its own Aion Labs tokenizer groupVery high
- Maker
- Aion Labs
- Label on the catalogue
- no family named
- What the measurement says
- its own Aion Labs tokenizer group
- Why very high
- an exact fingerprint shared with 1 model and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_13620098 · shared with 1 other model
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $3.00 / M tokens
- Entered the catalogue
- 23 September 2026
- Last tested
- 5 October 2026
AionLabs: Aion 3.5 Miniaion-labs/aion-3.5-miniits own Aion Labs tokenizer groupVery high
- Maker
- Aion Labs
- Label on the catalogue
- no family named
- What the measurement says
- its own Aion Labs tokenizer group
- Why very high
- an exact fingerprint shared with 1 model and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_13620098 · shared with 1 other model
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.70 / M tokens
- Entered the catalogue
- 23 September 2026
- Last tested
- 5 October 2026
- Verified finding
- Measured on 5 October 2026. The OpenRouter listing describes Aion 3.5 Mini as built on the GLM family of models, and Aion 3.0, which carries the same sentence, does measure with the GLM group on all 50 strings. Aion 3.5 Mini agrees with its sibling Aion 3.5 on…
AionLabs: Aion-2.0aion-labs/aion-2.0its own Aion Labs tokenizer groupModerate
- Maker
- Aion Labs
- Label on the catalogue
- no family named
- What the measurement says
- its own Aion Labs tokenizer group
- Why moderate
- an exact fingerprint match with one sibling only, and no second signal of a different kind yet
- Fingerprint
- tk_d3c6ae80 · shared with 1 other model
- Context
- 131,072
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.80 / M tokens
- Entered the catalogue
- 23 February 2026
- Last tested
- 3 September 2026
AionLabs: Aion-3.0aion-labs/aion-3.0builds on the Z.ai tokenizerModerate
- Maker
- Aion Labs
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Z.ai tokenizer
- Why moderate
- one template-boundary shift from the group rather than an exact signature member
- Fingerprint
- tk_c018f9f2 · unique so far
- Context
- 131,072
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $3.00 / M tokens
- Entered the catalogue
- 7 July 2026
- Last tested
- 3 September 2026
AionLabs: Aion-3.0-Miniaion-labs/aion-3.0-miniits own Aion Labs tokenizer groupModerate
- Maker
- Aion Labs
- Label on the catalogue
- no family named
- What the measurement says
- its own Aion Labs tokenizer group
- Why moderate
- an exact fingerprint match with one sibling only, and no second signal of a different kind yet
- Fingerprint
- tk_d3c6ae80 · shared with 1 other model
- Context
- 131,072
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.70 / M tokens
- Entered the catalogue
- 7 July 2026
- Last tested
- 3 September 2026
AionLabs: Aion-RP 1.0 (8B)aion-labs/aion-rp-llama-3.1-8bmeasured, matches nothing else yetMeasured
- Maker
- Aion Labs
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_559886c2 · unique so far
- Context
- 32,768
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.80 / M tokens
- Entered the catalogue
- 4 February 2025
- Last tested
- 3 September 2026
Amazon: Nova 2 Liteamazon/nova-2-lite-v1measured, matches nothing else yetMeasured
- Maker
- Amazon
- Label on the catalogue
- Nova
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_f04fd33e · unique so far
- Context
- 1,000,000
- Max output
- 65,535
- Modalities
- text+image+file+video->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 2 December 2025
- Last tested
- 21 August 2026
Amazon: Nova Premier 1.0amazon/nova-premier-v1measured, matches nothing else yetMeasured
- Maker
- Amazon
- Label on the catalogue
- Nova
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_79cd40d1 · unique so far
- Context
- 1,000,000
- Max output
- 32,000
- Modalities
- text+image->text
- Prompt price
- $2.50 / M tokens
- Entered the catalogue
- 31 October 2025
- Last tested
- 3 September 2026
Anthropic: Claude 3 Haikuanthropic/claude-3-haikuwithdrawn from the catalogue · measured as built on the Claude tokenizerHigh
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- withdrawn from the catalogue · measured as built on the Claude tokenizer
- Why high
- an exact fingerprint shared with 4 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_2d06485c · shared with 4 other models
- Context
- 200,000
- Max output
- 4,096
- Modalities
- text+image->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 13 March 2024
- Last tested
- 21 August 2026
Anthropic: Claude Haiku 5.5anthropic/claude-haiku-5.5builds on the Claude tokenizerHigh
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_310f7ba9 · shared with 1 other model
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 7 October 2026
- Last tested
- 8 October 2026
- Verified finding
- Measured on 8 October 2026, the day after it was listed. Claude Haiku 5.5 counts tokens exactly as Claude Sonnet 5.5 does on all 50 strings, and as Claude Opus 5.5, Opus 5 and Fable 5.1 do on the 49 they return (Anthropic's input filter stops the base64 string…
Anthropic: Claude Opus 4.7anthropic/claude-opus-4.7measured, matches nothing else yetMeasured
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- measured, matches nothing else yet
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $5.00 / M tokens
- Entered the catalogue
- 16 April 2026
- Last tested
- 3 September 2026
Anthropic: Claude Opus 5.5anthropic/claude-opus-5.5builds on the Claude tokenizerModerate
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with anthropic/claude-opus-5, short of a full exact signature
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $4.00 / M tokens
- Entered the catalogue
- 22 September 2026
- Last tested
- 5 October 2026
- Verified finding
- Opus 5.5 reads text with the same tokenizer as Opus 5 and Fable 5.1: all 49 measured strings return identical counts, measured on its release day. Its serving envelope matches Fable 5.1 in two further ways. The chat template adds 8 tokens to every request, aga…
Apodex: Apodex 1.1 Mini (free)apodex/apodex-1.1-mini:freebuilds on the Qwen tokenizerHigh
- Maker
- Apodex
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 5 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_06ffa14a · shared with 5 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 1 October 2026
- Last tested
- 5 October 2026
Arcee AI: Trinity Large Thinkingarcee-ai/trinity-large-thinkingmeasured, matches nothing else yetMeasured
- Maker
- Arcee AI
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_a8b24bf6 · unique so far
- Context
- 262,144
- Max output
- 80,000
- Modalities
- text->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 1 April 2026
- Last tested
- 21 August 2026
Baidu: ERNIE 4.5 VL 424B A47B baidu/ernie-4.5-vl-424b-a47bmeasured, matches nothing else yetMeasured
- Maker
- Baidu
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_41571a13 · unique so far
- Context
- 123,000
- Max output
- 16,000
- Modalities
- text+image->text
- Prompt price
- $0.42 / M tokens
- Entered the catalogue
- 30 June 2025
- Last tested
- 22 August 2026
ByteDance Seed: Seed 1.6bytedance-seed/seed-1.6its own ByteDance Seed tokenizer groupVery high
- Maker
- ByteDance Seed
- Label on the catalogue
- no family named
- What the measurement says
- its own ByteDance Seed tokenizer group
- Why very high
- an exact fingerprint shared with 5 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_25bf52d0 · shared with 5 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text+image+video->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 23 December 2025
- Last tested
- 21 August 2026
ByteDance Seed: Seed 1.6 Flashbytedance-seed/seed-1.6-flashits own ByteDance Seed tokenizer groupVery high
- Maker
- ByteDance Seed
- Label on the catalogue
- no family named
- What the measurement says
- its own ByteDance Seed tokenizer group
- Why very high
- an exact fingerprint shared with 5 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_25bf52d0 · shared with 5 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text+image+video->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 23 December 2025
- Last tested
- 21 August 2026
ByteDance Seed: Seed 2.1 Turbobytedance-seed/seed-2-1-turboits own ByteDance Seed tokenizer groupHigh
- Maker
- ByteDance Seed
- Label on the catalogue
- no family named
- What the measurement says
- its own ByteDance Seed tokenizer group
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_25bf52d0 · shared with 5 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image+video->text
- Prompt price
- $0.50 / M tokens
- Entered the catalogue
- 12 August 2026
- Last tested
- 5 October 2026
ByteDance Seed: Seed-2.0-Codebytedance-seed/seed-2.0-codeits own ByteDance Seed tokenizer groupHigh
- Maker
- ByteDance Seed
- Label on the catalogue
- no family named
- What the measurement says
- its own ByteDance Seed tokenizer group
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_25bf52d0 · shared with 5 other models
- Context
- 262,144
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $0.50 / M tokens
- Entered the catalogue
- 12 August 2026
- Last tested
- 22 August 2026
ByteDance Seed: Seed-2.0-Litebytedance-seed/seed-2.0-liteits own ByteDance Seed tokenizer groupHigh
- Maker
- ByteDance Seed
- Label on the catalogue
- no family named
- What the measurement says
- its own ByteDance Seed tokenizer group
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_25bf52d0 · shared with 5 other models
- Context
- 262,144
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 10 March 2026
- Last tested
- 21 August 2026
ByteDance Seed: Seed-2.0-Minibytedance-seed/seed-2.0-miniits own ByteDance Seed tokenizer groupHigh
- Maker
- ByteDance Seed
- Label on the catalogue
- no family named
- What the measurement says
- its own ByteDance Seed tokenizer group
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_25bf52d0 · shared with 5 other models
- Context
- 262,144
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 26 February 2026
- Last tested
- 21 August 2026
ByteDance: UI-TARS 7B bytedance/ui-tars-1.5-7bbuilds on the Qwen tokenizerHigh
- Maker
- Bytedance
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 128,000
- Max output
- 2,048
- Modalities
- text+image->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 22 July 2025
- Last tested
- 21 August 2026
Cohere: Command Acohere/command-aits own Cohere tokenizer groupHigh
- Maker
- Cohere
- Label on the catalogue
- no family named
- What the measurement says
- its own Cohere tokenizer group
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_2729e470 · shared with 1 other model
- Context
- 256,000
- Max output
- 8,192
- Modalities
- text->text
- Prompt price
- $2.50 / M tokens
- Entered the catalogue
- 13 March 2025
- Last tested
- 5 October 2026
DeepSeek: DeepSeek V3.2deepseek/deepseek-v3.2its own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 163,840
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $0.28 / M tokens
- Entered the catalogue
- 1 December 2025
- Last tested
- 21 August 2026
- Verified finding
- Re-measured on 5 October 2026 pinned to each of two providers. DigitalOcean reproduced the 21 August reading on every one of the 49 strings it answered cleanly. AtlasCloud, which served the August reading, now counts one string, the mathematical-alphabet chara…
DeepSeek: R1 Distill Llama 70Bdeepseek/deepseek-r1-distill-llama-70bwithdrawn from the catalogue · measured as its own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- Llama3
- What the measurement says
- withdrawn from the catalogue · measured as its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_869dc1cf · shared with 2 other models
- Context
- 8,192
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.80 / M tokens
- Entered the catalogue
- 23 January 2025
- Last tested
- 21 August 2026
- Verified finding
- The mismatch was traced to the serving layer, not the label. The model's own repository tokenizer, run over the same 50 strings offline, is Llama 3's exactly (50 of 50, vocabulary 128,256), so the catalogue's Llama3 tag is right about the weights. The endpoint…
Dots Studio: Dots3-Note Preview (free)dots-studio/dots-3-note-preview:freebuilds on the Qwen tokenizerHigh
- Maker
- Dots Studio
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 512,000
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 14 August 2026
- Last tested
- 21 August 2026
Fireworks: Ember-1fireworks/ember-1builds on the Moonshot AI tokenizerHigh
- Maker
- Fireworks
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Moonshot AI tokenizer
- Why high
- an exact fingerprint shared with 4 models
- Fingerprint
- tk_ea6068bf · shared with 4 other models
- Context
- 1,048,576
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $3.00 / M tokens
- Entered the catalogue
- 24 September 2026
- Last tested
- 5 October 2026
- Verified finding
- Listed on OpenRouter at 00:07 UTC on 24 September 2026 and measured the same day. Fireworks describes Ember-1 as a reasoning model from Fireworks Research built on Moonshot AI's Kimi K3, trained to reach Kimi K3's quality with about 40% fewer tokens. Its token…
Google: Gemini 2.5 Pro Preview 05-06google/gemini-2.5-pro-preview-05-06withdrawn from the catalogue · measured as built on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- withdrawn from the catalogue · measured as built on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,535
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $1.25 / M tokens
- Entered the catalogue
- 7 May 2025
- Last tested
- 3 September 2026
Google: Gemma 2 27Bgoogle/gemma-2-27b-itmeasured, matches nothing else yetMeasured
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_a73edaf2 · unique so far
- Context
- 8,192
- Max output
- 2,048
- Modalities
- text->text
- Prompt price
- $0.65 / M tokens
- Entered the catalogue
- 13 July 2024
- Last tested
- 3 September 2026
Google: Gemma 3n 4Bgoogle/gemma-3n-e4b-itwithdrawn from the catalogue · measured as built on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as built on the Gemini tokenizer
- Why high
- one template-boundary shift from a group whose members all declare the same family
- Fingerprint
- tk_a15512c7 · unique so far
- Context
- 32,768
- Max output
- –
- Modalities
- text->text
- Prompt price
- $0.06 / M tokens
- Entered the catalogue
- 20 May 2025
- Last tested
- 21 August 2026
Google: Lyria 3 Clip Previewgoogle/lyria-3-clip-previewmeasured, matches nothing else yetMeasured
- Maker
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image->text+audio
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 30 March 2026
- Last tested
- 3 September 2026
IBM: Granite 4.0 Microibm-granite/granite-4.0-h-microbuilds on the OpenAI cl100k tokenizerHigh
- Maker
- IBM
- Label on the catalogue
- no family named
- What the measurement says
- builds on the OpenAI cl100k tokenizer
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_ba918336 · shared with 5 other models
- Context
- 131,000
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.02 / M tokens
- Entered the catalogue
- 20 October 2025
- Last tested
- 21 August 2026
- Verified finding
- Granite 4.0's endpoint counts match OpenAI's open cl100k_base encoding on all 50 strings, and its repository declares the matching 100,352-entry vocabulary. Microsoft's Phi-4 measures identically, which is why two rival labs share one signature group: both ado…
IBM: Granite 4.1 8Bibm-granite/granite-4.1-8bwithdrawn from the catalogue · measured as built on the OpenAI cl100k tokenizerHigh
- Maker
- IBM
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as built on the OpenAI cl100k tokenizer
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_ba918336 · shared with 5 other models
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 30 April 2026
- Last tested
- 21 August 2026
- Verified finding
- Granite 4.1's endpoint counts match OpenAI's open cl100k_base encoding on all 50 strings, the same result as Granite 4.0 and Microsoft's Phi-4. The three share one signature group because both labs adopted the tokenizer OpenAI released as open source with tikt…
IBM: Granite 4.2 8Bibm-granite/granite-4.2-8bbuilds on the Inference Net tokenizerModerate
- Maker
- IBM
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Inference Net tokenizer
- Why moderate
- an exact fingerprint match with one sibling only, and no second signal of a different kind yet
- Fingerprint
- tk_b6cc7610 · shared with 1 other model
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.06 / M tokens
- Entered the catalogue
- 31 August 2026
- Last tested
- 3 September 2026
inclusionAI: Ling 3.0 Flashinclusionai/ling-3.0-flashits own InclusionAI tokenizer groupHigh
- Maker
- InclusionAI
- Label on the catalogue
- no family named
- What the measurement says
- its own InclusionAI tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_fc5a4585 · shared with 6 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.02 / M tokens
- Entered the catalogue
- 23 July 2026
- Last tested
- 21 August 2026
inclusionAI: Ling 3.0 Flash Fininclusionai/ling-3.0-flash-finits own InclusionAI tokenizer groupHigh
- Maker
- InclusionAI
- Label on the catalogue
- no family named
- What the measurement says
- its own InclusionAI tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_fc5a4585 · shared with 6 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.04 / M tokens
- Entered the catalogue
- 27 August 2026
- Last tested
- 3 September 2026
inclusionAI: Ling 3.0 Flash Sante (free)inclusionai/ling-3.0-flash-sante:freeits own InclusionAI tokenizer groupHigh
- Maker
- InclusionAI
- Label on the catalogue
- no family named
- What the measurement says
- its own InclusionAI tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_fc5a4585 · shared with 6 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 4 September 2026
- Last tested
- 5 October 2026
inclusionAI: Ling 3.0 Flash VLinclusionai/ling-3.0-flash-vlits own InclusionAI tokenizer groupHigh
- Maker
- InclusionAI
- Label on the catalogue
- no family named
- What the measurement says
- its own InclusionAI tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_fc5a4585 · shared with 6 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text+image+video->text
- Prompt price
- $0.02 / M tokens
- Entered the catalogue
- 10 September 2026
- Last tested
- 5 October 2026
Full recordShares with inclusionAI: Ling 3.0 Flash Sante (free)
inclusionAI: Ling-2.6-1Tinclusionai/ling-2.6-1twithdrawn from the catalogue · measured as its own InclusionAI tokenizer groupVery high
- Maker
- InclusionAI
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as its own InclusionAI tokenizer group
- Why very high
- an exact fingerprint shared with 6 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_fc5a4585 · shared with 6 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 23 April 2026
- Last tested
- 21 August 2026
inclusionAI: Ling-2.6-flashinclusionai/ling-2.6-flashwithdrawn from the catalogue · measured as its own InclusionAI tokenizer groupVery high
- Maker
- InclusionAI
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as its own InclusionAI tokenizer group
- Why very high
- an exact fingerprint shared with 6 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_fc5a4585 · shared with 6 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.01 / M tokens
- Entered the catalogue
- 21 April 2026
- Last tested
- 21 August 2026
inclusionAI: Ring-2.6-1Tinclusionai/ring-2.6-1twithdrawn from the catalogue · measured as its own InclusionAI tokenizer groupHigh
- Maker
- InclusionAI
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as its own InclusionAI tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_fc5a4585 · shared with 6 other models
- Context
- 262,144
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 8 May 2026
- Last tested
- 21 August 2026
Inference.net: Schematron V2 Smallinference-net/schematron-v2-smallbuilds on the Llama3 tokenizerHigh
- Maker
- Inference Net
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 8 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_6588ed5b · shared with 8 other models
- Context
- 128,000
- Max output
- 4,096
- Modalities
- text->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 12 September 2026
- Last tested
- 5 October 2026
Inference.net: Schematron V2 Turboinference-net/schematron-v2-turbobuilds on the IBM tokenizerModerate
- Maker
- Inference Net
- Label on the catalogue
- no family named
- What the measurement says
- builds on the IBM tokenizer
- Why moderate
- an exact fingerprint match with one sibling only, and no second signal of a different kind yet
- Fingerprint
- tk_b6cc7610 · shared with 1 other model
- Context
- 128,000
- Max output
- 8,192
- Modalities
- text->text
- Prompt price
- $0.03 / M tokens
- Entered the catalogue
- 12 September 2026
- Last tested
- 5 October 2026
Kwaipilot: KAT-Coder-Air V2.5kwaipilot/kat-coder-air-v2.5withdrawn from the catalogue · measured as built on the Qwen tokenizerHigh
- Maker
- Kwaipilot
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as built on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 5 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_06ffa14a · shared with 5 other models
- Context
- 256,000
- Max output
- 80,000
- Modalities
- text->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 10 July 2026
- Last tested
- 21 August 2026
Kwaipilot: KAT-Coder-Pro V2kwaipilot/kat-coder-pro-v2withdrawn from the catalogue · measured as built on the Qwen tokenizerHigh
- Maker
- Kwaipilot
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as built on the Qwen tokenizer
- Why high
- one template-boundary shift from a group whose members all declare the same family
- Fingerprint
- tk_e60ca14a · unique so far
- Context
- 262,144
- Max output
- 144,000
- Modalities
- text->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 27 March 2026
- Last tested
- 21 August 2026
LiquidAI: LFM2.5-2.6B (free)liquid/lfm-2.5-2.6b:freemeasured, matches nothing else yetMeasured
- Maker
- Liquid AI
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_2c236574 · unique so far
- Context
- 65,536
- Max output
- 8,192
- Modalities
- text->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 11 August 2026
- Last tested
- 21 August 2026
Meituan: LongCat 2.0meituan/longcat-2.0measured, matches nothing else yetMeasured
- Maker
- Meituan
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_c670dac9 · unique so far
- Context
- 1,048,756
- Max output
- 262,144
- Modalities
- text->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 20 July 2026
- Last tested
- 21 August 2026
Meta: Llama Guard 4 12Bmeta-llama/llama-guard-4-12bmeasured, matches nothing else yetMeasured
- Maker
- Meta
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_985608d3 · unique so far
- Context
- 163,840
- Max output
- 16,384
- Modalities
- text+image->text
- Prompt price
- $0.18 / M tokens
- Entered the catalogue
- 30 April 2025
- Last tested
- 21 August 2026
Meta: Muse Glimmer 30Bmeta/muse-glimmer-30bbuilds on the Llama4 tokenizerHigh
- Maker
- Meta
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Llama4 tokenizer
- Why high
- one template-boundary shift from a group whose members all declare the same family
- Fingerprint
- tk_20b9d005 · unique so far
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.35 / M tokens
- Entered the catalogue
- 9 August 2026
- Last tested
- 21 August 2026
Microsoft: Phi 4microsoft/phi-4builds on the OpenAI cl100k tokenizerHigh
- Maker
- Microsoft
- Label on the catalogue
- no family named
- What the measurement says
- builds on the OpenAI cl100k tokenizer
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_ba918336 · shared with 5 other models
- Context
- 16,384
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 10 January 2025
- Last tested
- 21 August 2026
- Verified finding
- Phi-4's repository tokenizer, run over the 50 strings offline, matches OpenAI's open cl100k_base encoding on every string (vocabulary 100,352, the padded cl100k table). IBM's Granite 4.0 and 4.1 declare the same vocabulary size and their endpoints return ident…
MiniMax: MiniMax M1minimax/minimax-m1its own MiniMax tokenizer groupHigh
- Maker
- MiniMax
- Label on the catalogue
- no family named
- What the measurement says
- its own MiniMax tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_9f84b5fb · shared with 8 other models
- Context
- 1,000,000
- Max output
- 40,000
- Modalities
- text->text
- Prompt price
- $0.55 / M tokens
- Entered the catalogue
- 17 June 2025
- Last tested
- 22 August 2026
MiniMax: MiniMax M2minimax/minimax-m2its own MiniMax tokenizer groupHigh
- Maker
- MiniMax
- Label on the catalogue
- no family named
- What the measurement says
- its own MiniMax tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_9f84b5fb · shared with 8 other models
- Context
- 204,800
- Max output
- 176,947
- Modalities
- text->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 23 October 2025
- Last tested
- 21 August 2026
MiniMax: MiniMax M2-herminimax/minimax-m2-herits own MiniMax tokenizer groupHigh
- Maker
- MiniMax
- Label on the catalogue
- no family named
- What the measurement says
- its own MiniMax tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_9f84b5fb · shared with 8 other models
- Context
- 65,536
- Max output
- 2,048
- Modalities
- text->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 23 January 2026
- Last tested
- 21 August 2026
MiniMax: MiniMax M2.1minimax/minimax-m2.1its own MiniMax tokenizer groupHigh
- Maker
- MiniMax
- Label on the catalogue
- no family named
- What the measurement says
- its own MiniMax tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_9f84b5fb · shared with 8 other models
- Context
- 204,800
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 23 December 2025
- Last tested
- 21 August 2026
MiniMax: MiniMax M2.5minimax/minimax-m2.5its own MiniMax tokenizer groupHigh
- Maker
- MiniMax
- Label on the catalogue
- no family named
- What the measurement says
- its own MiniMax tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_9f84b5fb · shared with 8 other models
- Context
- 204,800
- Max output
- 128,000
- Modalities
- text->text
- Prompt price
- $0.27 / M tokens
- Entered the catalogue
- 12 February 2026
- Last tested
- 21 August 2026
MiniMax: MiniMax M2.7minimax/minimax-m2.7its own MiniMax tokenizer groupHigh
- Maker
- MiniMax
- Label on the catalogue
- no family named
- What the measurement says
- its own MiniMax tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_9f84b5fb · shared with 8 other models
- Context
- 204,800
- Max output
- 176,947
- Modalities
- text->text
- Prompt price
- $0.21 / M tokens
- Entered the catalogue
- 18 March 2026
- Last tested
- 5 October 2026
MiniMax: MiniMax M3minimax/minimax-m3its own MiniMax tokenizer groupHigh
- Maker
- MiniMax
- Label on the catalogue
- no family named
- What the measurement says
- its own MiniMax tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_9f84b5fb · shared with 8 other models
- Context
- 1,048,576
- Max output
- 512,000
- Modalities
- text+image+video->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 31 May 2026
- Last tested
- 5 October 2026
MiniMax: MiniMax-01minimax/minimax-01its own MiniMax tokenizer groupHigh
- Maker
- MiniMax
- Label on the catalogue
- no family named
- What the measurement says
- its own MiniMax tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_9f84b5fb · shared with 8 other models
- Context
- 1,000,192
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 15 January 2025
- Last tested
- 21 August 2026
Mistral: Mistral Large 4mistralai/mistral-large-4-0measured, matches nothing else yetMeasured
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_e242e95c · unique so far
- Context
- 1,048,576
- Max output
- 262,144
- Modalities
- text+image->text
- Prompt price
- $0.68 / M tokens
- Entered the catalogue
- 6 October 2026
- Last tested
- 8 October 2026
- Verified finding
- Measured on 8 October 2026, two days after it was listed, and measured again the same day pinned to Mistral's own endpoint: both readings agree on all 50 strings. Mistral Large 4 runs a tokenizer no other model on record uses, Mistral's own included. It agrees…
Mistral: Mistral Small 3mistralai/mistral-small-24b-instruct-2501measured, matches nothing else yetMeasured
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_ddc193e9 · shared with 2 other models
- Context
- 32,768
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 30 January 2025
- Last tested
- 21 August 2026
MoonshotAI: Kimi K2 0711moonshotai/kimi-k2measured, matches nothing else yetMeasured
- Maker
- Moonshot AI
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_2e353ea4 · unique so far
- Context
- 131,072
- Max output
- 98,304
- Modalities
- text->text
- Prompt price
- $0.57 / M tokens
- Entered the catalogue
- 11 July 2025
- Last tested
- 22 August 2026
MoonshotAI: Kimi K2 0905moonshotai/kimi-k2-0905its own Moonshot AI tokenizer groupModerate
- Maker
- Moonshot AI
- Label on the catalogue
- no family named
- What the measurement says
- its own Moonshot AI tokenizer group
- Why moderate
- near-exact agreement (49 of 50 strings) with moonshotai/kimi-k3, short of a full exact signature
- Fingerprint
- tk_f14a0fdd · unique so far
- Context
- 262,144
- Max output
- 98,304
- Modalities
- text->text
- Prompt price
- $0.60 / M tokens
- Entered the catalogue
- 4 September 2025
- Last tested
- 3 September 2026
MoonshotAI: Kimi K2 Thinkingmoonshotai/kimi-k2-thinkingits own Moonshot AI tokenizer groupHigh
- Maker
- Moonshot AI
- Label on the catalogue
- no family named
- What the measurement says
- its own Moonshot AI tokenizer group
- Why high
- an exact fingerprint shared with 4 models
- Fingerprint
- tk_ea6068bf · shared with 4 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.60 / M tokens
- Entered the catalogue
- 6 November 2025
- Last tested
- 3 September 2026
MoonshotAI: Kimi K2.5moonshotai/kimi-k2.5its own Moonshot AI tokenizer groupModerate
- Maker
- Moonshot AI
- Label on the catalogue
- no family named
- What the measurement says
- its own Moonshot AI tokenizer group
- Why moderate
- near-exact agreement (48 of 48 strings) with moonshotai/kimi-k3, short of a full exact signature
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.45 / M tokens
- Entered the catalogue
- 27 January 2026
- Last tested
- 3 September 2026
MoonshotAI: Kimi K2.6moonshotai/kimi-k2.6its own Moonshot AI tokenizer groupHigh
- Maker
- Moonshot AI
- Label on the catalogue
- no family named
- What the measurement says
- its own Moonshot AI tokenizer group
- Why high
- an exact fingerprint shared with 4 models
- Fingerprint
- tk_ea6068bf · shared with 4 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.95 / M tokens
- Entered the catalogue
- 20 April 2026
- Last tested
- 5 October 2026
MoonshotAI: Kimi K2.7 Codemoonshotai/kimi-k2.7-codeits own Moonshot AI tokenizer groupHigh
- Maker
- Moonshot AI
- Label on the catalogue
- no family named
- What the measurement says
- its own Moonshot AI tokenizer group
- Why high
- an exact fingerprint shared with 4 models
- Fingerprint
- tk_ea6068bf · shared with 4 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.67 / M tokens
- Entered the catalogue
- 12 June 2026
- Last tested
- 5 October 2026
MoonshotAI: Kimi K3moonshotai/kimi-k3its own Moonshot AI tokenizer groupHigh
- Maker
- Moonshot AI
- Label on the catalogue
- no family named
- What the measurement says
- its own Moonshot AI tokenizer group
- Why high
- an exact fingerprint shared with 4 models
- Fingerprint
- tk_ea6068bf · shared with 4 other models
- Context
- 1,048,576
- Max output
- undeclared
- Modalities
- text+image+video->text
- Prompt price
- $0.67 / M tokens
- Entered the catalogue
- 16 July 2026
- Last tested
- 5 October 2026
- Verified finding
- Re-measured on 24 September 2026 on a single provider, Fireworks, as the control for Fireworks' Ember-1. The 21 August sweep read this model while OpenRouter spread the calls across seven providers, and eight of its rows came back exactly 3 tokens low, the dif…
Morph: Morph V3 Fastmorph/morph-v3-fastbuilds on the Qwen tokenizerHigh
- Maker
- Morph
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 81,920
- Max output
- 38,000
- Modalities
- text->text
- Prompt price
- $0.80 / M tokens
- Entered the catalogue
- 7 July 2025
- Last tested
- 3 September 2026
Morph: Morph V3 Largemorph/morph-v3-largebuilds on the Qwen tokenizerHigh
- Maker
- Morph
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_c5c0a8ed · shared with 2 other models
- Context
- 262,144
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $0.90 / M tokens
- Entered the catalogue
- 7 July 2025
- Last tested
- 3 September 2026
Nex AGI: Nex-N2-Mininex-agi/nex-n2-miniwithdrawn from the catalogue · measured as built on the Qwen tokenizerHigh
- Maker
- Nex Agi
- Label on the catalogue
- Qwen3
- What the measurement says
- withdrawn from the catalogue · measured as built on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.02 / M tokens
- Entered the catalogue
- 24 June 2026
- Last tested
- 21 August 2026
Nex AGI: Nex-N2-Pronex-agi/nex-n2-prowithdrawn from the catalogue · measured as built on the Qwen tokenizerHigh
- Maker
- Nex Agi
- Label on the catalogue
- Qwen3
- What the measurement says
- withdrawn from the catalogue · measured as built on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 8 June 2026
- Last tested
- 21 August 2026
Nous: Hermes 4 405Bnousresearch/hermes-4-405bbuilds on the Llama3 tokenizerVery high
- Maker
- Nous Research
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Llama3 tokenizer
- Why very high
- an exact fingerprint shared with 4 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_47615f2d · shared with 4 other models
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $1.00 / M tokens
- Entered the catalogue
- 26 August 2025
- Last tested
- 3 September 2026
Nous: Hermes 4 70Bnousresearch/hermes-4-70bwithdrawn from the catalogue · measured as built on the Llama3 tokenizerVery high
- Maker
- Nous Research
- Label on the catalogue
- Llama3
- What the measurement says
- withdrawn from the catalogue · measured as built on the Llama3 tokenizer
- Why very high
- an exact fingerprint shared with 4 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_47615f2d · shared with 4 other models
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.13 / M tokens
- Entered the catalogue
- 26 August 2025
- Last tested
- 21 August 2026
NVIDIA: Nemotron 3 Nano 30B A3Bnvidia/nemotron-3-nano-30b-a3bbuilds on the Mistral tokenizerHigh
- Maker
- NVIDIA
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 14 December 2025
- Last tested
- 21 August 2026
NVIDIA: Nemotron 3 Nano Omni (free)nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:freebuilds on the Mistral tokenizerHigh
- Maker
- NVIDIA
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- one template-boundary shift from a group whose members all declare the same family
- Fingerprint
- tk_e0b4c23e · unique so far
- Context
- 256,000
- Max output
- 65,536
- Modalities
- text+image+audio+video->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 28 April 2026
- Last tested
- 21 August 2026
NVIDIA: Nemotron 3 Supernvidia/nemotron-3-super-120b-a12bbuilds on the Mistral tokenizerHigh
- Maker
- NVIDIA
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.08 / M tokens
- Entered the catalogue
- 11 March 2026
- Last tested
- 21 August 2026
NVIDIA: Nemotron 3 Ultranvidia/nemotron-3-ultra-550b-a55bbuilds on the Mistral tokenizerHigh
- Maker
- NVIDIA
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.50 / M tokens
- Entered the catalogue
- 4 June 2026
- Last tested
- 5 October 2026
NVIDIA: Nemotron 3.5 Content Safetynvidia/nemotron-3.5-content-safetyits own NVIDIA tokenizer groupLow
- Maker
- NVIDIA
- Label on the catalogue
- no family named
- What the measurement says
- its own NVIDIA tokenizer group
- Why low
- partial agreement (43 of 50 strings) with nvidia/nemotron-3.5-content-safety:free only
- Fingerprint
- tk_0d9ebf63 · unique so far
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 4 June 2026
- Last tested
- 5 October 2026
NVIDIA: Nemotron 3.5 Content Safety (free)nvidia/nemotron-3.5-content-safety:freeits own NVIDIA tokenizer groupLow
- Maker
- NVIDIA
- Label on the catalogue
- no family named
- What the measurement says
- its own NVIDIA tokenizer group
- Why low
- inherited from nvidia/nemotron-3.5-content-safety: a serving variant shares its base model's tokenizer
- Fingerprint
- tk_9a840805 · unique so far
- Variant of
- NVIDIA: Nemotron 3.5 Content Safety
- Context
- 128,000
- Max output
- 8,192
- Modalities
- text+image->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 4 June 2026
- Last tested
- 21 August 2026
NVIDIA: Nemotron 3.5 Lightningnvidia/nemotron-3.5-lightningbuilds on the Mistral tokenizerHigh
- Maker
- NVIDIA
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.06 / M tokens
- Entered the catalogue
- 11 August 2026
- Last tested
- 21 August 2026
NVIDIA: Nemotron Nano 12B 2 VL (free)nvidia/nemotron-nano-12b-v2-vl:freewithdrawn from the catalogue · measured as built on the Mistral tokenizerLow
- Maker
- NVIDIA
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as built on the Mistral tokenizer
- Why low
- partial agreement (40 of 47 strings) with nvidia/nemotron-nano-9b-v2:free only
- Context
- 128,000
- Max output
- 128,000
- Modalities
- text+image+video->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 28 October 2025
- Last tested
- 21 August 2026
NVIDIA: Nemotron Nano 9B V2 (free)nvidia/nemotron-nano-9b-v2:freewithdrawn from the catalogue · measured as built on the Mistral tokenizerHigh
- Maker
- NVIDIA
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as built on the Mistral tokenizer
- Why high
- one template-boundary shift from a group whose members all declare the same family
- Fingerprint
- tk_b0a747db · unique so far
- Context
- 128,000
- Max output
- –
- Modalities
- text->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 5 September 2025
- Last tested
- 21 August 2026
OpenAI: GPT-3.5 Turboopenai/gpt-3.5-turbobuilds on the OpenAI cl100k tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the OpenAI cl100k tokenizer
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_ba918336 · shared with 5 other models
- Context
- 16,385
- Max output
- 4,096
- Modalities
- text->text
- Prompt price
- $0.50 / M tokens
- Entered the catalogue
- 28 May 2023
- Last tested
- 22 August 2026
OpenAI: GPT-3.5 Turbo 16kopenai/gpt-3.5-turbo-16kbuilds on the OpenAI cl100k tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the OpenAI cl100k tokenizer
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_ba918336 · shared with 5 other models
- Context
- 16,385
- Max output
- 4,096
- Modalities
- text->text
- Prompt price
- $3.00 / M tokens
- Entered the catalogue
- 28 August 2023
- Last tested
- 5 October 2026
OpenAI: GPT-4 Turboopenai/gpt-4-turbobuilds on the OpenAI cl100k tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the OpenAI cl100k tokenizer
- Why high
- an exact fingerprint shared with 5 models
- Fingerprint
- tk_ba918336 · shared with 5 other models
- Context
- 128,000
- Max output
- 4,096
- Modalities
- text+image->text
- Prompt price
- $10.0 / M tokens
- Entered the catalogue
- 9 April 2024
- Last tested
- 5 October 2026
Ox Alphastealth/ox-alphawithdrawn from the catalogue · measured as built on the Z.ai GLM tokenizerVery high
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as built on the Z.ai GLM tokenizer
- Why very high
- Z.ai confirmed Ox Alpha as GLM-5.3-Flash on 26 August 2026, five days after the measurement matched it to GLM-5.3 on all fifty strings
- Fingerprint
- tk_133b086c · shared with 8 other models
- Context
- 1,048,576
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 20 August 2026
- Last tested
- 21 August 2026
Paretounbiased/paretomeasured, matches nothing else yetMeasured
- Maker
- Unbiased
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_7d8ce940 · unique so far
- Context
- 262,144
- Max output
- 131,072
- Modalities
- text+image->text
- Prompt price
- $2.50 / M tokens
- Entered the catalogue
- 17 September 2026
- Last tested
- 5 October 2026
- Verified finding
- Measured on 5 October 2026 through the endpoint Unbiased serves itself, with a fixed overhead of zero. The prompt counts it reports sit at about half of what any tokenizer on record produces, and on the strings that separate tokenizers they fall to one: a Chin…
Pareto 26.10 Previewunbiased/pareto-26.10-previewbuilds on the Llama3 tokenizerModerate
- Maker
- Unbiased
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Llama3 tokenizer
- Why moderate
- near-exact agreement (48 of 49 strings) with meta-llama/llama-3.2-3b-instruct, short of a full exact signature
- Context
- 1,048,576
- Max output
- 131,072
- Modalities
- text+image->text
- Prompt price
- $0.80 / M tokens
- Entered the catalogue
- 1 October 2026
- Last tested
- 5 October 2026
Perceptron: Perceptron Mk1perceptron/perceptron-mk1builds on the Qwen tokenizerHigh
- Maker
- Perceptron
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_c5c0a8ed · shared with 2 other models
- Context
- 32,768
- Max output
- 8,192
- Modalities
- text+image+video->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 12 May 2026
- Last tested
- 21 August 2026
Perceptron: Perceptron Mk1.5perceptron/perceptron-mk1.5builds on the Qwen tokenizerHigh
- Maker
- Perceptron
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_c5c0a8ed · shared with 2 other models
- Context
- 36,864
- Max output
- 8,192
- Modalities
- text+image+audio+video->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 25 September 2026
- Last tested
- 5 October 2026
Perplexity: Sonarperplexity/sonarbuilds on the Llama3 tokenizerHigh
- Maker
- Perplexity
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 8 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_6588ed5b · shared with 8 other models
- Context
- 127,072
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $1.00 / M tokens
- Entered the catalogue
- 27 January 2025
- Last tested
- 5 October 2026
Perplexity: Sonar Properplexity/sonar-probuilds on the Llama3 tokenizerHigh
- Maker
- Perplexity
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 8 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_6588ed5b · shared with 8 other models
- Context
- 200,000
- Max output
- 8,000
- Modalities
- text+image->text
- Prompt price
- $3.00 / M tokens
- Entered the catalogue
- 7 March 2025
- Last tested
- 3 September 2026
Poolside: Laguna S 2.1poolside/laguna-s-2.1measured, matches nothing else yetMeasured
- Maker
- Poolside
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_cd5bb5f6 · unique so far
- Context
- 1,048,576
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $0.09 / M tokens
- Entered the catalogue
- 21 July 2026
- Last tested
- 21 August 2026
Poolside: Laguna XS 2.1poolside/laguna-xs-2.1measured, matches nothing else yetMeasured
- Maker
- Poolside
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_08973fc0 · unique so far
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.06 / M tokens
- Entered the catalogue
- 2 July 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.8 27Bqwen/qwen3.8-27bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 5 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_06ffa14a · shared with 5 other models
- Context
- 1,000,000
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $0.42 / M tokens
- Entered the catalogue
- 14 August 2026
- Last tested
- 5 October 2026
- Verified finding
- Re-measured on 5 October 2026 on four providers one at a time: Reka, CoreWeave, Parasail and Chutes. The 21 August reading was taken while OpenRouter spread the calls across eight providers, and five of its rows (English prose, Hindi, the emoji sequence, the m…
Qwen: Qwen3.8 Maxqwen/qwen3.8-maxwithdrawn from the catalogue · measured as built on the Qwen tokenizerVery high
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- withdrawn from the catalogue · measured as built on the Qwen tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 3 August 2026
- Last tested
- 3 September 2026
Reka Edgerekaai/reka-edgemeasured, matches nothing else yetMeasured
- Maker
- Rekaai
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_9f6f8410 · unique so far
- Context
- 16,384
- Max output
- undeclared
- Modalities
- text+image+video->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 20 March 2026
- Last tested
- 21 August 2026
Reka Flash 3rekaai/reka-flash-3measured, matches nothing else yetMeasured
- Maker
- Rekaai
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Context
- 65,536
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 12 March 2025
- Last tested
- 21 August 2026
Relace: Relace Searchrelace/relace-searchbuilds on the Qwen tokenizerHigh
- Maker
- Relace
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 256,000
- Max output
- 128,000
- Modalities
- text->text
- Prompt price
- $1.00 / M tokens
- Entered the catalogue
- 8 December 2025
- Last tested
- 3 September 2026
Space Bunny Alphastealth/space-bunny-alphalikely MiniMax familyHigh
- Maker
- Undisclosed (stealth)
- Label on the catalogue
- no family named
- What the measurement says
- likely MiniMax family
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_9f84b5fb · shared with 8 other models
- Context
- 1,000,000
- Max output
- 524,288
- Modalities
- text+image+video->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 23 September 2026
- Last tested
- 23 September 2026
- Verified finding
- Listed anonymously at 14:48 UTC on 23 September 2026 and measured the same day. Its token counts are MiniMax's on all 50 strings: every one of the eight MiniMax models in the dataset, from MiniMax-01 to M3, returns the same 50 counts, against 29 of 50 for the …
StepFun: Step 3.5 Flashstepfun/step-3.5-flashbuilds on the DeepSeek tokenizerHigh
- Maker
- StepFun
- Label on the catalogue
- no family named
- What the measurement says
- builds on the DeepSeek tokenizer
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 262,144
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 29 January 2026
- Last tested
- 21 August 2026
- Verified finding
- StepFun's own Step-3 repository ships a tokenizer behaviourally identical to DeepSeek-V3's: run over the 50 strings offline the two agree on every one, and both declare a 128,815-entry vocabulary. The endpoint's counts match as well, so this model sitting in t…
StepFun: Step 3.7 Flashstepfun/step-3.7-flashbuilds on the DeepSeek tokenizerHigh
- Maker
- StepFun
- Label on the catalogue
- no family named
- What the measurement says
- builds on the DeepSeek tokenizer
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 262,144
- Max output
- 230,400
- Modalities
- text+image+video->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 28 May 2026
- Last tested
- 5 October 2026
- Verified finding
- As with Step 3.5 Flash: StepFun's own Step-3 repository ships a tokenizer behaviourally identical to DeepSeek-V3's over all 50 strings, and this endpoint counts the same way. Its place in the DeepSeek signature group is genuine lineage reuse by StepFun.…
StepFun: Step 5 Previewstepfun/step-5-previewbuilds on the DeepSeek tokenizerHigh
- Maker
- StepFun
- Label on the catalogue
- no family named
- What the measurement says
- builds on the DeepSeek tokenizer
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 1,000,000
- Max output
- 64,000
- Modalities
- text+image+video->text
- Prompt price
- $1.00 / M tokens
- Entered the catalogue
- 8 October 2026
- Last tested
- 8 October 2026
- Verified finding
- Measured on 8 October 2026. Step 5 Preview counts tokens exactly as DeepSeek V4 Pro, V4 Flash and the rest of DeepSeek's current line do on all 50 strings, as StepFun's Step 3.5 Flash and Step 3.7 Flash already did. StepFun's flagship keeps DeepSeek's vocabula…
Tencent: Hunyuan A13B Instructtencent/hunyuan-a13b-instructits own Tencent tokenizer groupModerate
- Maker
- Tencent
- Label on the catalogue
- no family named
- What the measurement says
- its own Tencent tokenizer group
- Why moderate
- near-exact agreement (49 of 50 strings) with tencent/hy-mt2-7b, short of a full exact signature
- Fingerprint
- tk_856fd527 · unique so far
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.14 / M tokens
- Entered the catalogue
- 8 July 2025
- Last tested
- 21 August 2026
Tencent: Hy-MT2-1.8Btencent/hy-mt2-1.8bits own Tencent tokenizer groupHigh
- Maker
- Tencent
- Label on the catalogue
- no family named
- What the measurement says
- its own Tencent tokenizer group
- Why high
- an exact fingerprint shared with 3 models
- Fingerprint
- tk_2b541395 · shared with 3 other models
- Context
- 8,192
- Max output
- 4,096
- Modalities
- text->text
- Prompt price
- $0.04 / M tokens
- Entered the catalogue
- 20 August 2026
- Last tested
- 21 August 2026
Tencent: Hy-MT2-30B-A3Btencent/hy-mt2-30b-a3bits own Tencent tokenizer groupHigh
- Maker
- Tencent
- Label on the catalogue
- no family named
- What the measurement says
- its own Tencent tokenizer group
- Why high
- an exact fingerprint shared with 3 models
- Fingerprint
- tk_2b541395 · shared with 3 other models
- Context
- 8,192
- Max output
- 4,096
- Modalities
- text->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 20 August 2026
- Last tested
- 21 August 2026
Tencent: Hy-MT2-7Btencent/hy-mt2-7bits own Tencent tokenizer groupModerate
- Maker
- Tencent
- Label on the catalogue
- no family named
- What the measurement says
- its own Tencent tokenizer group
- Why moderate
- near-exact agreement (49 of 50 strings) with tencent/hunyuan-a13b-instruct, short of a full exact signature
- Fingerprint
- tk_70a22c4f · unique so far
- Context
- 8,192
- Max output
- 4,096
- Modalities
- text->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 19 August 2026
- Last tested
- 3 September 2026
Tencent: Hy3tencent/hy3its own Tencent tokenizer groupHigh
- Maker
- Tencent
- Label on the catalogue
- no family named
- What the measurement says
- its own Tencent tokenizer group
- Why high
- an exact fingerprint shared with 3 models
- Fingerprint
- tk_2b541395 · shared with 3 other models
- Context
- 262,144
- Max output
- 128,000
- Modalities
- text->text
- Prompt price
- $0.08 / M tokens
- Entered the catalogue
- 6 July 2026
- Last tested
- 21 August 2026
Tencent: Hy3 previewtencent/hy3-previewits own Tencent tokenizer groupHigh
- Maker
- Tencent
- Label on the catalogue
- no family named
- What the measurement says
- its own Tencent tokenizer group
- Why high
- an exact fingerprint shared with 3 models
- Fingerprint
- tk_2b541395 · shared with 3 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.18 / M tokens
- Entered the catalogue
- 22 April 2026
- Last tested
- 21 August 2026
Tencent: Hy4 previewtencent/hy4-previewmeasured, matches nothing else yetMeasured
- Maker
- Tencent
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_5e7942ae · unique so far
- Context
- 1,048,576
- Max output
- 64,000
- Modalities
- text->text
- Prompt price
- $0.75 / M tokens
- Entered the catalogue
- 28 August 2026
- Last tested
- 5 October 2026
TheDrummer: Cydonia 24B V4.1thedrummer/cydonia-24b-v4.1builds on the Mistral tokenizerModerate
- Maker
- Thedrummer
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Mistral tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with mistralai/mistral-small-24b-instruct-2501, short of a full exact signature
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 27 September 2025
- Last tested
- 21 August 2026
TheDrummer: Skyfall 36B V2thedrummer/skyfall-36b-v2builds on the Mistral tokenizerHigh
- Maker
- Thedrummer
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_ddc193e9 · shared with 2 other models
- Context
- 32,768
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.55 / M tokens
- Entered the catalogue
- 10 March 2025
- Last tested
- 22 August 2026
TheDrummer: UnslopNemo 12Bthedrummer/unslopnemo-12bmeasured, matches nothing else yetMeasured
- Maker
- Thedrummer
- Label on the catalogue
- Mistral
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_f573b76b · unique so far
- Context
- 1,024,000
- Max output
- 819,200
- Modalities
- text->text
- Prompt price
- $0.40 / M tokens
- Entered the catalogue
- 8 November 2024
- Last tested
- 22 August 2026
Thinking Machines: Inklingthinkingmachines/inklingbuilds on the GPT tokenizerVery high
- Maker
- Thinking Machines
- Label on the catalogue
- no family named
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 524,288
- Max output
- 262,144
- Modalities
- text+image+audio->text
- Prompt price
- $0.95 / M tokens
- Entered the catalogue
- 17 July 2026
- Last tested
- 3 September 2026
Thinking Machines: Inkling Smallthinkingmachines/inkling-smallbuilds on the GPT tokenizerVery high
- Maker
- Thinking Machines
- Label on the catalogue
- no family named
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 524,288
- Max output
- 262,144
- Modalities
- text+image+audio->text
- Prompt price
- $0.45 / M tokens
- Entered the catalogue
- 30 July 2026
- Last tested
- 22 August 2026
TypeSafe: Jev 1.13typesafe/jev-1.13measured, matches nothing else yetMeasured
- Maker
- TypeSafe
- Label on the catalogue
- no family named
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_5e398719 · unique so far
- Context
- 32,000
- Max output
- undeclared
- Modalities
- text->decisions
- Prompt price
- $0.04 / M tokens
- Entered the catalogue
- 18 September 2026
- Last tested
- 18 September 2026
Union Alphastealth/union-alphawithdrawn from the catalogue · measured as built on the Pareto (multi-model blend) tokenizerVery high
- Maker
- Circuit and Chisel
- Label on the catalogue
- no family named
- What the measurement says
- withdrawn from the catalogue · measured as built on the Pareto (multi-model blend) tokenizer
- Why very high
- the operator confirmed Union Alpha as Pareto 26.9 on 17 September 2026, 18 hours after the YFarmX post named it; the blend carries the Llama 3, Qwen and GLM tokenizers the measurements had read
- Context
- 262,144
- Max output
- 131,072
- Modalities
- text+image->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 16 September 2026
- Last tested
- 17 September 2026
- Verified finding
- Listed anonymously on 16 September 2026 and measured the same day. This endpoint returns TWO different token counts for a byte-identical request, and they describe different tokenizers, which is why early readings of it disagreed. A differential test settles w…
Upstage: Solar Mini 4upstage/solar-mini4its own Upstage tokenizer groupVery high
- Maker
- Upstage
- Label on the catalogue
- no family named
- What the measurement says
- its own Upstage tokenizer group
- Why very high
- an exact fingerprint shared with 2 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_865d84ce · shared with 2 other models
- Context
- 524,288
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 23 September 2026
- Last tested
- 5 October 2026
Upstage: Solar Pro 3upstage/solar-pro-3its own Upstage tokenizer groupHigh
- Maker
- Upstage
- Label on the catalogue
- no family named
- What the measurement says
- its own Upstage tokenizer group
- Why high
- an exact fingerprint shared with 2 models
- Fingerprint
- tk_865d84ce · shared with 2 other models
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 27 January 2026
- Last tested
- 21 August 2026
Upstage: Solar Pro 4upstage/solar-pro4its own Upstage tokenizer groupVery high
- Maker
- Upstage
- Label on the catalogue
- no family named
- What the measurement says
- its own Upstage tokenizer group
- Why very high
- an exact fingerprint shared with 2 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_865d84ce · shared with 2 other models
- Context
- 524,288
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $0.09 / M tokens
- Entered the catalogue
- 10 August 2026
- Last tested
- 21 August 2026
Venice: Uncensoredcognitivecomputations/dolphin-mistral-24b-venice-editionbuilds on the Mistral tokenizerHigh
- Maker
- Cognitivecomputations
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_ddc193e9 · shared with 2 other models
- Context
- 128,000
- Max output
- 8,192
- Modalities
- text->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 9 July 2025
- Last tested
- 21 August 2026
WizardLM-2 8x22Bmicrosoft/wizardlm-2-8x22bmeasured, matches nothing else yetMeasured
- Maker
- Microsoft
- Label on the catalogue
- Mistral
- What the measurement says
- measured, matches nothing else yet
- Fingerprint
- tk_a49eb75d · unique so far
- Context
- 65,535
- Max output
- 8,000
- Modalities
- text->text
- Prompt price
- $0.62 / M tokens
- Entered the catalogue
- 16 April 2024
- Last tested
- 22 August 2026
Writer: Palmyra X5writer/palmyra-x5builds on the Qwen tokenizerHigh
- Maker
- Writer
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- one template-boundary shift from a group whose members all declare the same family
- Fingerprint
- tk_920d45ce · unique so far
- Context
- 1,040,000
- Max output
- 8,192
- Modalities
- text->text
- Prompt price
- $0.60 / M tokens
- Entered the catalogue
- 21 January 2026
- Last tested
- 3 September 2026
Xiaomi: MiMo-V2.5xiaomi/mimo-v2.5builds on the Qwen tokenizerHigh
- Maker
- Xiaomi
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 1,050,000
- Max output
- 131,072
- Modalities
- text+image+audio+video->text
- Prompt price
- $0.14 / M tokens
- Entered the catalogue
- 22 April 2026
- Last tested
- 5 October 2026
Xiaomi: MiMo-V2.5-Proxiaomi/mimo-v2.5-probuilds on the Qwen tokenizerHigh
- Maker
- Xiaomi
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 1,050,000
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $0.43 / M tokens
- Entered the catalogue
- 22 April 2026
- Last tested
- 5 October 2026
- Verified finding
- Re-measured on 5 October 2026 on GMICloud. Every one of the 50 strings came back exactly two tokens above the 22 August reading, and the new reading matches MiMo V2.5 on all 50. That is the artefact the 4 September note on this record predicted: the August bas…
Xiaomi: MiMo-V2.6-Flashxiaomi/mimo-v2.6-flashbuilds on the Qwen tokenizerHigh
- Maker
- Xiaomi
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 1,050,000
- Max output
- 131,072
- Modalities
- text+image+audio+video->text
- Prompt price
- $0.14 / M tokens
- Entered the catalogue
- 21 September 2026
- Last tested
- 5 October 2026
Xiaomi: MiMo-V2.6-Proxiaomi/mimo-v2.6-probuilds on the Qwen tokenizerHigh
- Maker
- Xiaomi
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 1,050,000
- Max output
- 131,072
- Modalities
- text+image+audio+video->text
- Prompt price
- $0.43 / M tokens
- Entered the catalogue
- 21 September 2026
- Last tested
- 5 October 2026
Xiaomi: MiMo-V2.6-Pro-UltraSpeedxiaomi/mimo-v2.6-pro-ultraspeedbuilds on the Qwen tokenizerHigh
- Maker
- Xiaomi
- Label on the catalogue
- no family named
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 1,048,576
- Max output
- 131,072
- Modalities
- text+image+audio+video->text
- Prompt price
- $4.35 / M tokens
- Entered the catalogue
- 21 September 2026
- Last tested
- 5 October 2026
Z.ai: GLM 4.5z-ai/glm-4.5its own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_3eaaf3e0 · shared with 6 other models
- Context
- 131,072
- Max output
- 98,304
- Modalities
- text->text
- Prompt price
- $0.60 / M tokens
- Entered the catalogue
- 25 July 2025
- Last tested
- 3 September 2026
Z.ai: GLM 4.5 Airz-ai/glm-4.5-airits own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_3eaaf3e0 · shared with 6 other models
- Context
- 131,072
- Max output
- 98,304
- Modalities
- text->text
- Prompt price
- $0.13 / M tokens
- Entered the catalogue
- 25 July 2025
- Last tested
- 21 August 2026
Z.ai: GLM 4.5Vz-ai/glm-4.5vits own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_3eaaf3e0 · shared with 6 other models
- Context
- 65,536
- Max output
- 16,384
- Modalities
- text+image->text
- Prompt price
- $0.60 / M tokens
- Entered the catalogue
- 11 August 2025
- Last tested
- 22 August 2026
Z.ai: GLM 4.6z-ai/glm-4.6its own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_3eaaf3e0 · shared with 6 other models
- Context
- 204,800
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.43 / M tokens
- Entered the catalogue
- 30 September 2025
- Last tested
- 22 August 2026
Z.ai: GLM 4.6Vz-ai/glm-4.6vits own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_3eaaf3e0 · shared with 6 other models
- Context
- 131,072
- Max output
- 32,768
- Modalities
- text+image+video->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 8 December 2025
- Last tested
- 21 August 2026
Z.ai: GLM 4.7z-ai/glm-4.7its own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_3eaaf3e0 · shared with 6 other models
- Context
- 204,800
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $0.60 / M tokens
- Entered the catalogue
- 22 December 2025
- Last tested
- 22 August 2026
Z.ai: GLM 4.7 Flashz-ai/glm-4.7-flashits own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_133b086c · shared with 8 other models
- Context
- 200,000
- Max output
- 117,964
- Modalities
- text->text
- Prompt price
- $0.06 / M tokens
- Entered the catalogue
- 19 January 2026
- Last tested
- 21 August 2026
Z.ai: GLM 5z-ai/glm-5its own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_133b086c · shared with 8 other models
- Context
- 204,800
- Max output
- 128,000
- Modalities
- text->text
- Prompt price
- $0.60 / M tokens
- Entered the catalogue
- 11 February 2026
- Last tested
- 5 October 2026
Z.ai: GLM 5 Turboz-ai/glm-5-turboits own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_133b086c · shared with 8 other models
- Context
- 202,752
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $1.20 / M tokens
- Entered the catalogue
- 15 March 2026
- Last tested
- 3 September 2026
Z.ai: GLM 5.1z-ai/glm-5.1its own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_133b086c · shared with 8 other models
- Context
- 204,800
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $1.40 / M tokens
- Entered the catalogue
- 7 April 2026
- Last tested
- 5 October 2026
- Verified finding
- Re-measured on 5 October 2026 pinned to each of two providers. Friendli, which served the 3 September reading, reproduced it on all 50 strings. AtlasCloud, which now also serves this model, reports a fixed overhead of 12 tokens against Friendli's 5 and counts …
Z.ai: GLM 5.2z-ai/glm-5.2its own Z.ai tokenizer groupModerate
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why moderate
- one template-boundary shift from the group rather than an exact signature member
- Fingerprint
- tk_1d05586d · unique so far
- Context
- 1,048,576
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $0.02 / M tokens
- Entered the catalogue
- 16 June 2026
- Last tested
- 21 August 2026
Z.ai: GLM 5.3z-ai/glm-5.3its own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_133b086c · shared with 8 other models
- Context
- 1,048,576
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 18 August 2026
- Last tested
- 5 October 2026
Z.ai: GLM 5.3 Flashz-ai/glm-5.3-flashits own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_133b086c · shared with 8 other models
- Context
- 1,048,576
- Max output
- 943,717
- Modalities
- text+image+video->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 26 August 2026
- Last tested
- 5 October 2026
Z.ai: GLM 5.3 FlashXz-ai/glm-5.3-flashxits own Z.ai tokenizer groupVery high
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why very high
- an exact fingerprint shared with 8 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_133b086c · shared with 8 other models
- Context
- 1,048,576
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $0.37 / M tokens
- Entered the catalogue
- 18 September 2026
- Last tested
- 5 October 2026
Z.ai: GLM 5.3 Primez-ai/glm-5.3-primeits own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 8 models
- Fingerprint
- tk_133b086c · shared with 8 other models
- Context
- 1,000,000
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $2.80 / M tokens
- Entered the catalogue
- 23 September 2026
- Last tested
- 5 October 2026
Z.ai: GLM 5V Turboz-ai/glm-5v-turboits own Z.ai tokenizer groupHigh
- Maker
- Z.ai
- Label on the catalogue
- no family named
- What the measurement says
- its own Z.ai tokenizer group
- Why high
- an exact fingerprint shared with 6 models
- Fingerprint
- tk_3eaaf3e0 · shared with 6 other models
- Context
- 202,752
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $1.20 / M tokens
- Entered the catalogue
- 1 April 2026
- Last tested
- 3 September 2026
Amazon: Nova Lite 1.0amazon/nova-lite-v1builds on the Nova tokenizerVery high
- Maker
- Amazon
- Label on the catalogue
- Nova
- What the measurement says
- builds on the Nova tokenizer
- Why very high
- an exact fingerprint shared with 2 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_76e424dd · shared with 2 other models
- Context
- 300,000
- Max output
- 5,120
- Modalities
- text+image->text
- Prompt price
- $0.06 / M tokens
- Entered the catalogue
- 5 December 2024
- Last tested
- 21 August 2026
Amazon: Nova Micro 1.0amazon/nova-micro-v1builds on the Nova tokenizerHigh
- Maker
- Amazon
- Label on the catalogue
- Nova
- What the measurement says
- builds on the Nova tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_76e424dd · shared with 2 other models
- Context
- 128,000
- Max output
- 5,120
- Modalities
- text->text
- Prompt price
- $0.04 / M tokens
- Entered the catalogue
- 5 December 2024
- Last tested
- 21 August 2026
Amazon: Nova Pro 1.0amazon/nova-pro-v1builds on the Nova tokenizerVery high
- Maker
- Amazon
- Label on the catalogue
- Nova
- What the measurement says
- builds on the Nova tokenizer
- Why very high
- an exact fingerprint shared with 2 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_76e424dd · shared with 2 other models
- Context
- 300,000
- Max output
- 5,120
- Modalities
- text+image->text
- Prompt price
- $0.80 / M tokens
- Entered the catalogue
- 5 December 2024
- Last tested
- 3 September 2026
Anthropic: Claude Fable 5anthropic/claude-fable-5builds on the Claude tokenizerModerate
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why moderate
- near-exact agreement (44 of 44 strings) with anthropic/claude-sonnet-5, short of a full exact signature
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $10.0 / M tokens
- Entered the catalogue
- 9 June 2026
- Last tested
- 3 September 2026
Anthropic: Claude Fable 5.1anthropic/claude-fable-5.1builds on the Claude tokenizerModerate
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with anthropic/claude-opus-5, short of a full exact signature
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $10.0 / M tokens
- Entered the catalogue
- 1 September 2026
- Last tested
- 5 October 2026
Anthropic: Claude Haiku 4.5anthropic/claude-haiku-4.5builds on the Claude tokenizerHigh
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why high
- an exact fingerprint shared with 4 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_2d06485c · shared with 4 other models
- Context
- 200,000
- Max output
- 64,000
- Modalities
- text+image+file->text
- Prompt price
- $1.00 / M tokens
- Entered the catalogue
- 15 October 2025
- Last tested
- 5 October 2026
Anthropic: Claude Opus 4.1anthropic/claude-opus-4.1builds on the Claude tokenizerModerate
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with anthropic/claude-3-haiku, short of a full exact signature
- Context
- 200,000
- Max output
- 32,000
- Modalities
- text+image+file->text
- Prompt price
- $15.0 / M tokens
- Entered the catalogue
- 5 August 2025
- Last tested
- 5 October 2026
Anthropic: Claude Opus 4.5anthropic/claude-opus-4.5builds on the Claude tokenizerHigh
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why high
- an exact fingerprint shared with 4 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_2d06485c · shared with 4 other models
- Context
- 200,000
- Max output
- 64,000
- Modalities
- text+image+file->text
- Prompt price
- $5.00 / M tokens
- Entered the catalogue
- 24 November 2025
- Last tested
- 5 October 2026
Anthropic: Claude Opus 4.6anthropic/claude-opus-4.6builds on the Claude tokenizerModerate
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with anthropic/claude-3-haiku, short of a full exact signature
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $5.00 / M tokens
- Entered the catalogue
- 4 February 2026
- Last tested
- 5 October 2026
Anthropic: Claude Opus 4.8anthropic/claude-opus-4.8builds on the Claude tokenizerModerate
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with anthropic/claude-sonnet-5, short of a full exact signature
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $5.00 / M tokens
- Entered the catalogue
- 27 May 2026
- Last tested
- 5 October 2026
Anthropic: Claude Opus 5anthropic/claude-opus-5builds on the Claude tokenizerModerate
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with anthropic/claude-fable-5.1, short of a full exact signature
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $5.00 / M tokens
- Entered the catalogue
- 24 July 2026
- Last tested
- 5 October 2026
Anthropic: Claude Sonnet 4anthropic/claude-sonnet-4builds on the Claude tokenizerHigh
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why high
- an exact fingerprint shared with 4 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_2d06485c · shared with 4 other models
- Context
- 200,000
- Max output
- 64,000
- Modalities
- text+image+file->text
- Prompt price
- $3.00 / M tokens
- Entered the catalogue
- 22 May 2025
- Last tested
- 5 October 2026
Anthropic: Claude Sonnet 4.5anthropic/claude-sonnet-4.5builds on the Claude tokenizerHigh
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why high
- an exact fingerprint shared with 4 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_2d06485c · shared with 4 other models
- Context
- 1,000,000
- Max output
- 64,000
- Modalities
- text+image+file->text
- Prompt price
- $3.00 / M tokens
- Entered the catalogue
- 29 September 2025
- Last tested
- 5 October 2026
Anthropic: Claude Sonnet 4.6anthropic/claude-sonnet-4.6builds on the Claude tokenizerModerate
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with anthropic/claude-3-haiku, short of a full exact signature
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $3.00 / M tokens
- Entered the catalogue
- 17 February 2026
- Last tested
- 3 September 2026
Anthropic: Claude Sonnet 5anthropic/claude-sonnet-5builds on the Claude tokenizerModerate
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with anthropic/claude-opus-4.8, short of a full exact signature
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 30 June 2026
- Last tested
- 5 October 2026
Anthropic: Claude Sonnet 5.5anthropic/claude-sonnet-5.5builds on the Claude tokenizerHigh
- Maker
- Anthropic
- Label on the catalogue
- Claude
- What the measurement says
- builds on the Claude tokenizer
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_310f7ba9 · shared with 1 other model
- Context
- 1,000,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 28 September 2026
- Last tested
- 5 October 2026
Cohere: Command A+cohere/command-a-plusits own Cohere tokenizer groupHigh
- Maker
- Cohere
- Label on the catalogue
- Cohere
- What the measurement says
- its own Cohere tokenizer group
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_caaaa531 · shared with 1 other model
- Context
- 192,000
- Max output
- 64,000
- Modalities
- text+image->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 22 September 2026
- Last tested
- 5 October 2026
Cohere: Command R (08-2024)cohere/command-r-08-2024its own Cohere tokenizer groupModerate
- Maker
- Cohere
- Label on the catalogue
- Cohere
- What the measurement says
- its own Cohere tokenizer group
- Why moderate
- near-exact agreement (49 of 49 strings) with cohere/command-r7b-12-2024, short of a full exact signature
- Context
- 128,000
- Max output
- 4,000
- Modalities
- text->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 30 August 2024
- Last tested
- 21 August 2026
Cohere: Command R+ (08-2024)cohere/command-r-plus-08-2024its own Cohere tokenizer groupModerate
- Maker
- Cohere
- Label on the catalogue
- Cohere
- What the measurement says
- its own Cohere tokenizer group
- Why moderate
- near-exact agreement (49 of 49 strings) with cohere/command-r7b-12-2024, short of a full exact signature
- Context
- 128,000
- Max output
- 4,000
- Modalities
- text->text
- Prompt price
- $2.50 / M tokens
- Entered the catalogue
- 30 August 2024
- Last tested
- 5 October 2026
Cohere: Command R7B (12-2024)cohere/command-r7b-12-2024its own Cohere tokenizer groupModerate
- Maker
- Cohere
- Label on the catalogue
- Cohere
- What the measurement says
- its own Cohere tokenizer group
- Why moderate
- an exact fingerprint match with one sibling only, and no second signal of a different kind yet
- Fingerprint
- tk_2729e470 · shared with 1 other model
- Context
- 128,000
- Max output
- 4,000
- Modalities
- text->text
- Prompt price
- $0.04 / M tokens
- Entered the catalogue
- 14 December 2024
- Last tested
- 21 August 2026
Cohere: North Mini Code (free)cohere/north-mini-code:freeits own Cohere tokenizer groupHigh
- Maker
- Cohere
- Label on the catalogue
- Cohere
- What the measurement says
- its own Cohere tokenizer group
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_caaaa531 · shared with 1 other model
- Context
- 256,000
- Max output
- 64,000
- Modalities
- text->text
- Prompt price
- $0 / M tokens
- Entered the catalogue
- 17 June 2026
- Last tested
- 21 August 2026
DeepSeek: DeepSeek V3deepseek/deepseek-chatits own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_869dc1cf · shared with 2 other models
- Context
- 163,840
- Max output
- 16,000
- Modalities
- text->text
- Prompt price
- $0.26 / M tokens
- Entered the catalogue
- 26 December 2024
- Last tested
- 5 October 2026
DeepSeek: DeepSeek V3 0324deepseek/deepseek-chat-v3-0324its own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 163,840
- Max output
- 115,200
- Modalities
- text->text
- Prompt price
- $0.29 / M tokens
- Entered the catalogue
- 24 March 2025
- Last tested
- 21 August 2026
DeepSeek: DeepSeek V3.1deepseek/deepseek-chat-v3.1its own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_869dc1cf · shared with 2 other models
- Context
- 163,840
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 21 August 2025
- Last tested
- 21 August 2026
DeepSeek: DeepSeek V3.1 Terminusdeepseek/deepseek-v3.1-terminusits own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 163,840
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.27 / M tokens
- Entered the catalogue
- 22 September 2025
- Last tested
- 21 August 2026
DeepSeek: DeepSeek V3.2 Expdeepseek/deepseek-v3.2-expits own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 163,840
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.27 / M tokens
- Entered the catalogue
- 29 September 2025
- Last tested
- 21 August 2026
DeepSeek: DeepSeek V4 Flash 0423deepseek/deepseek-v4-flashits own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 1,048,576
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.03 / M tokens
- Entered the catalogue
- 24 April 2026
- Last tested
- 5 October 2026
DeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731its own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 1,048,576
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.02 / M tokens
- Entered the catalogue
- 31 July 2026
- Last tested
- 21 August 2026
DeepSeek: DeepSeek V4 Flash Vision Expdeepseek/deepseek-v4-flash-vision-expits own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 1,048,576
- Max output
- 262,144
- Modalities
- text+image->text
- Prompt price
- $0.22 / M tokens
- Entered the catalogue
- 21 August 2026
- Last tested
- 5 October 2026
DeepSeek: DeepSeek V4 Pro 0423deepseek/deepseek-v4-proits own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 1,048,576
- Max output
- 384,000
- Modalities
- text->text
- Prompt price
- $0.21 / M tokens
- Entered the catalogue
- 24 April 2026
- Last tested
- 5 October 2026
DeepSeek: DeepSeek V4 Pro 0813deepseek/deepseek-v4-pro-0813its own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 1,048,576
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.40 / M tokens
- Entered the catalogue
- 12 August 2026
- Last tested
- 21 August 2026
DeepSeek: DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flashits own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 1,048,576
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 10 September 2026
- Last tested
- 5 October 2026
Full recordShares with DeepSeek: DeepSeek V4 Flash Vision Exp
DeepSeek: R1deepseek/deepseek-r1its own DeepSeek tokenizer groupModerate
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why moderate
- near-exact agreement (47 of 50 strings) with deepseek/deepseek-r1-distill-llama-70b, short of a full exact signature
- Fingerprint
- tk_d6c0b2be · unique so far
- Context
- 64,000
- Max output
- 16,000
- Modalities
- text->text
- Prompt price
- $0.70 / M tokens
- Entered the catalogue
- 20 January 2025
- Last tested
- 5 October 2026
DeepSeek: R1 0528deepseek/deepseek-r1-0528its own DeepSeek tokenizer groupHigh
- Maker
- DeepSeek
- Label on the catalogue
- DeepSeek
- What the measurement says
- its own DeepSeek tokenizer group
- Why high
- an exact fingerprint shared with 13 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_73c4f0dc · shared with 13 other models
- Context
- 163,840
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.50 / M tokens
- Entered the catalogue
- 28 May 2025
- Last tested
- 3 September 2026
Google: Gemini 2.5 Flashgoogle/gemini-2.5-flashbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,535
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 17 June 2025
- Last tested
- 5 October 2026
Google: Gemini 2.5 Flash Litegoogle/gemini-2.5-flash-litebuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,535
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 22 July 2025
- Last tested
- 21 August 2026
Google: Gemini 2.5 Progoogle/gemini-2.5-probuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $1.25 / M tokens
- Entered the catalogue
- 17 June 2025
- Last tested
- 5 October 2026
Google: Gemini 2.5 Pro Preview 06-05google/gemini-2.5-pro-previewbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio->text
- Prompt price
- $1.25 / M tokens
- Entered the catalogue
- 5 June 2025
- Last tested
- 3 September 2026
Google: Gemini 3 Flash Previewgoogle/gemini-3-flash-previewbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $0.50 / M tokens
- Entered the catalogue
- 17 December 2025
- Last tested
- 22 August 2026
Google: Gemini 3.1 Flash Litegoogle/gemini-3.1-flash-litebuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 7 May 2026
- Last tested
- 21 August 2026
Google: Gemini 3.1 Flash Lite Previewgoogle/gemini-3.1-flash-lite-previewbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 3 March 2026
- Last tested
- 21 August 2026
Google: Gemini 3.1 Pro Previewgoogle/gemini-3.1-pro-previewbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 19 February 2026
- Last tested
- 5 October 2026
Google: Gemini 3.1 Pro Preview Custom Toolsgoogle/gemini-3.1-pro-preview-customtoolsbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 25 February 2026
- Last tested
- 3 September 2026
Google: Gemini 3.5 Flashgoogle/gemini-3.5-flashbuilds on the Gemini tokenizerVery high
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $1.50 / M tokens
- Entered the catalogue
- 19 May 2026
- Last tested
- 5 October 2026
Google: Gemini 3.5 Flash Litegoogle/gemini-3.5-flash-litebuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 21 July 2026
- Last tested
- 21 August 2026
Google: Gemini 3.6 Flashgoogle/gemini-3.6-flashbuilds on the Gemini tokenizerVery high
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $0.75 / M tokens
- Entered the catalogue
- 21 July 2026
- Last tested
- 5 October 2026
Google: Gemini 3.7 Flashgoogle/gemini-3.7-flashbuilds on the Gemini tokenizerVery high
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $0.75 / M tokens
- Entered the catalogue
- 13 August 2026
- Last tested
- 5 October 2026
Google: Gemini 3.8 Flashgoogle/gemini-3.8-flashbuilds on the Gemini tokenizerVery high
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 1,048,576
- Max output
- 65,536
- Modalities
- text+image+file+audio+video->text
- Prompt price
- $0.75 / M tokens
- Entered the catalogue
- 2 September 2026
- Last tested
- 5 October 2026
Google: Gemma 3 12Bgoogle/gemma-3-12b-itbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text+image->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 13 March 2025
- Last tested
- 21 August 2026
Google: Gemma 3 27Bgoogle/gemma-3-27b-itbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.08 / M tokens
- Entered the catalogue
- 12 March 2025
- Last tested
- 21 August 2026
Google: Gemma 3 4Bgoogle/gemma-3-4b-itbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text+image->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 13 March 2025
- Last tested
- 22 August 2026
Google: Gemma 4 26B A4B google/gemma-4-26b-a4b-itbuilds on the Gemma tokenizerHigh
- Maker
- Label on the catalogue
- Gemma
- What the measurement says
- builds on the Gemma tokenizer
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_ba03436b · shared with 1 other model
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image+video->text
- Prompt price
- $0.09 / M tokens
- Entered the catalogue
- 3 April 2026
- Last tested
- 21 August 2026
Google: Gemma 4 31Bgoogle/gemma-4-31b-itbuilds on the Gemma tokenizerHigh
- Maker
- Label on the catalogue
- Gemma
- What the measurement says
- builds on the Gemma tokenizer
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_ba03436b · shared with 1 other model
- Context
- 262,144
- Max output
- 16,384
- Modalities
- text+image+video->text
- Prompt price
- $0.09 / M tokens
- Entered the catalogue
- 2 April 2026
- Last tested
- 5 October 2026
Google: Nano Banana (Gemini 2.5 Flash Image)google/gemini-2.5-flash-imagebuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 32,768
- Max output
- 8,192
- Modalities
- text+image->text+image
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 7 October 2025
- Last tested
- 3 September 2026
Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)google/gemini-3.1-flash-image-previewbuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 65,536
- Max output
- undeclared
- Modalities
- text+image->text+image
- Prompt price
- $0.50 / M tokens
- Entered the catalogue
- 26 February 2026
- Last tested
- 22 August 2026
Google: Nano Banana 2 (Gemini 3.1 Flash Image)google/gemini-3.1-flash-imagebuilds on the Gemini tokenizerModerate
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why moderate
- near-exact agreement (44 of 44 strings) with google/gemini-3.7-flash, short of a full exact signature
- Context
- 131,072
- Max output
- 32,768
- Modalities
- text+image->text+image
- Prompt price
- $0.50 / M tokens
- Entered the catalogue
- 18 June 2026
- Last tested
- 22 August 2026
Google: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)google/gemini-3.1-flash-lite-imagebuilds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 65,536
- Max output
- undeclared
- Modalities
- text+image->text+image
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 30 June 2026
- Last tested
- 3 September 2026
Google: Nano Banana 2.1google/gemini-nano-banana-2.1builds on the Gemini tokenizerHigh
- Maker
- Label on the catalogue
- Gemini
- What the measurement says
- builds on the Gemini tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59ca250f · shared with 21 other models
- Context
- 65,536
- Max output
- undeclared
- Modalities
- text+image->text+image
- Prompt price
- $1.50 / M tokens
- Entered the catalogue
- 6 October 2026
- Last tested
- 8 October 2026
Magnum v4 72Banthracite-org/magnum-v4-72bbuilds on the Qwen tokenizerHigh
- Maker
- Anthracite Org
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 32,768
- Max output
- 4,096
- Modalities
- text->text
- Prompt price
- $2.50 / M tokens
- Entered the catalogue
- 22 October 2024
- Last tested
- 3 September 2026
Mancer: Weaver (alpha)mancer/weaverbuilds on the Llama2 tokenizerHigh
- Maker
- Mancer
- Label on the catalogue
- Llama2
- What the measurement says
- builds on the Llama2 tokenizer
- Why high
- one template-boundary shift from a group whose members all declare the same family
- Fingerprint
- tk_03ecdf53 · unique so far
- Context
- 8,000
- Max output
- 6,000
- Modalities
- text->text
- Prompt price
- $0.40 / M tokens
- Entered the catalogue
- 2 August 2023
- Last tested
- 21 August 2026
Meta: Llama 3.1 70B Instructmeta-llama/llama-3.1-70b-instructbuilds on the Llama3 tokenizerHigh
- Maker
- Meta
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 8 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_6588ed5b · shared with 8 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.40 / M tokens
- Entered the catalogue
- 23 July 2024
- Last tested
- 21 August 2026
Meta: Llama 3.1 8B Instructmeta-llama/llama-3.1-8b-instructbuilds on the Llama3 tokenizerHigh
- Maker
- Meta
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 8 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_6588ed5b · shared with 8 other models
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 23 July 2024
- Last tested
- 21 August 2026
Meta: Llama 3.2 1B Instructmeta-llama/llama-3.2-1b-instructbuilds on the Llama3 tokenizerHigh
- Maker
- Meta
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 4 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_47615f2d · shared with 4 other models
- Context
- 60,000
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.03 / M tokens
- Entered the catalogue
- 25 September 2024
- Last tested
- 21 August 2026
Meta: Llama 3.2 3B Instructmeta-llama/llama-3.2-3b-instructbuilds on the Llama3 tokenizerHigh
- Maker
- Meta
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 8 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_6588ed5b · shared with 8 other models
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 25 September 2024
- Last tested
- 21 August 2026
Meta: Llama 3.3 70B Instructmeta-llama/llama-3.3-70b-instructbuilds on the Llama3 tokenizerHigh
- Maker
- Meta
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 8 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_6588ed5b · shared with 8 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.22 / M tokens
- Entered the catalogue
- 6 December 2024
- Last tested
- 5 October 2026
Meta: Llama 4 Maverickmeta-llama/llama-4-maverickbuilds on the Llama4 tokenizerHigh
- Maker
- Meta
- Label on the catalogue
- Llama4
- What the measurement says
- builds on the Llama4 tokenizer
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59e2520c · shared with 1 other model
- Context
- 1,048,576
- Max output
- 16,384
- Modalities
- text+image->text
- Prompt price
- $0.19 / M tokens
- Entered the catalogue
- 5 April 2025
- Last tested
- 5 October 2026
Meta: Llama 4 Scoutmeta-llama/llama-4-scoutbuilds on the Llama4 tokenizerHigh
- Maker
- Meta
- Label on the catalogue
- Llama4
- What the measurement says
- builds on the Llama4 tokenizer
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_59e2520c · shared with 1 other model
- Context
- 1,310,720
- Max output
- 16,384
- Modalities
- text+image->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 5 April 2025
- Last tested
- 5 October 2026
Mistral Largemistralai/mistral-largebuilds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 128,000
- Max output
- 102,400
- Modalities
- text+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 26 February 2024
- Last tested
- 3 September 2026
Mistral Large 2407mistralai/mistral-large-2407builds on the Mistral tokenizerVery high
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 131,072
- Max output
- 104,857
- Modalities
- text+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 19 November 2024
- Last tested
- 3 September 2026
Mistral: Codestral 2508mistralai/codestral-2508builds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 256,000
- Max output
- 204,800
- Modalities
- text+file->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 1 August 2025
- Last tested
- 21 August 2026
Mistral: Devstral 2 2512mistralai/devstral-2512builds on the Mistral tokenizerVery high
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- 209,715
- Modalities
- text+file->text
- Prompt price
- $0.40 / M tokens
- Entered the catalogue
- 9 December 2025
- Last tested
- 3 September 2026
Mistral: Ministral 3 14B 2512mistralai/ministral-14b-2512builds on the Mistral tokenizerVery high
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- 209,715
- Modalities
- text+image->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 2 December 2025
- Last tested
- 21 August 2026
Mistral: Ministral 3 3B 2512mistralai/ministral-3b-2512builds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 131,072
- Max output
- 104,857
- Modalities
- text+image->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 2 December 2025
- Last tested
- 21 August 2026
Mistral: Ministral 3 8B 2512mistralai/ministral-8b-2512builds on the Mistral tokenizerVery high
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- 209,715
- Modalities
- text+image->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 2 December 2025
- Last tested
- 21 August 2026
Mistral: Mistral Large 3 2512mistralai/mistral-large-2512builds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- 209,715
- Modalities
- text+image+file->text
- Prompt price
- $0.50 / M tokens
- Entered the catalogue
- 1 December 2025
- Last tested
- 22 August 2026
Mistral: Mistral Medium 3mistralai/mistral-medium-3builds on the Mistral tokenizerVery high
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 131,072
- Max output
- 104,857
- Modalities
- text+image+file->text
- Prompt price
- $0.40 / M tokens
- Entered the catalogue
- 7 May 2025
- Last tested
- 22 August 2026
Mistral: Mistral Medium 3.1mistralai/mistral-medium-3.1builds on the Mistral tokenizerVery high
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 131,072
- Max output
- 104,857
- Modalities
- text+image+file->text
- Prompt price
- $0.40 / M tokens
- Entered the catalogue
- 13 August 2025
- Last tested
- 22 August 2026
Mistral: Mistral Medium 3.5mistralai/mistral-medium-3-5builds on the Mistral tokenizerVery high
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- 209,715
- Modalities
- text+image+file->text
- Prompt price
- $1.50 / M tokens
- Entered the catalogue
- 30 April 2026
- Last tested
- 5 October 2026
Mistral: Mistral Nemomistralai/mistral-nemobuilds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.02 / M tokens
- Entered the catalogue
- 19 July 2024
- Last tested
- 21 August 2026
Mistral: Mistral Small 3.1 24Bmistralai/mistral-small-3.1-24b-instructbuilds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 128,000
- Max output
- 102,400
- Modalities
- text+image->text
- Prompt price
- $0.35 / M tokens
- Entered the catalogue
- 17 March 2025
- Last tested
- 21 August 2026
Mistral: Mistral Small 3.2 24Bmistralai/mistral-small-3.2-24b-instructbuilds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 256,000
- Max output
- 16,384
- Modalities
- text+image->text
- Prompt price
- $0.09 / M tokens
- Entered the catalogue
- 20 June 2025
- Last tested
- 21 August 2026
Mistral: Mistral Small 4mistralai/mistral-small-2603builds on the Mistral tokenizerVery high
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 262,144
- Max output
- 209,715
- Modalities
- text+image->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 16 March 2026
- Last tested
- 5 October 2026
Mistral: Mixtral 8x22B Instructmistralai/mixtral-8x22b-instructbuilds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 65,536
- Max output
- 52,428
- Modalities
- text+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 17 April 2024
- Last tested
- 3 September 2026
Mistral: Sabamistralai/mistral-sababuilds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 32,768
- Max output
- 26,214
- Modalities
- text+file->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 17 February 2025
- Last tested
- 21 August 2026
Mistral: Voxtral Small 24B 2507mistralai/voxtral-small-24b-2507builds on the Mistral tokenizerHigh
- Maker
- Mistral AI
- Label on the catalogue
- Mistral
- What the measurement says
- builds on the Mistral tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_0da8a999 · shared with 21 other models
- Context
- 32,768
- Max output
- 26,214
- Modalities
- text+file+audio->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 30 October 2025
- Last tested
- 21 August 2026
MythoMax 13Bgryphe/mythomax-l2-13bbuilds on the Llama2 tokenizerHigh
- Maker
- Gryphe
- Label on the catalogue
- Llama2
- What the measurement says
- builds on the Llama2 tokenizer
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5348f9f6 · shared with 1 other model
- Context
- 8,192
- Max output
- 3,686
- Modalities
- text->text
- Prompt price
- $0.08 / M tokens
- Entered the catalogue
- 2 July 2023
- Last tested
- 21 August 2026
Nex AGI: Nex-N2.5-Mininex-agi/nex-n2.5-minibuilds on the Qwen tokenizerHigh
- Maker
- Nex Agi
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.02 / M tokens
- Entered the catalogue
- 8 September 2026
- Last tested
- 5 October 2026
Nex AGI: Nex-N2.5-Pronex-agi/nex-n2.5-probuilds on the Qwen tokenizerHigh
- Maker
- Nex Agi
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 5 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_06ffa14a · shared with 5 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 8 September 2026
- Last tested
- 5 October 2026
Nous: Hermes 3 405B Instructnousresearch/hermes-3-llama-3.1-405bbuilds on the Llama3 tokenizerHigh
- Maker
- Nous Research
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 4 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_47615f2d · shared with 4 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $1.00 / M tokens
- Entered the catalogue
- 16 August 2024
- Last tested
- 21 August 2026
Nous: Hermes 3 70B Instructnousresearch/hermes-3-llama-3.1-70bbuilds on the Llama3 tokenizerHigh
- Maker
- Nous Research
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 4 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_47615f2d · shared with 4 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.70 / M tokens
- Entered the catalogue
- 18 August 2024
- Last tested
- 21 August 2026
OpenAI: GPT Chat Latestopenai/gpt-chat-latestbuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $5.00 / M tokens
- Entered the catalogue
- 5 May 2026
- Last tested
- 5 October 2026
OpenAI: GPT-3.5 Turbo (older v0613)openai/gpt-3.5-turbo-0613builds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 4,095
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $1.00 / M tokens
- Entered the catalogue
- 25 January 2024
- Last tested
- 3 September 2026
OpenAI: GPT-4.1openai/gpt-4.1builds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,047,576
- Max output
- 32,768
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 14 April 2025
- Last tested
- 5 October 2026
OpenAI: GPT-4.1 Miniopenai/gpt-4.1-minibuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,047,576
- Max output
- 32,768
- Modalities
- text+image+file->text
- Prompt price
- $0.40 / M tokens
- Entered the catalogue
- 14 April 2025
- Last tested
- 22 August 2026
OpenAI: GPT-4.1 Nanoopenai/gpt-4.1-nanobuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,047,576
- Max output
- 32,768
- Modalities
- text+image+file->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 14 April 2025
- Last tested
- 21 August 2026
OpenAI: GPT-4oopenai/gpt-4obuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 128,000
- Max output
- 16,384
- Modalities
- text+image+file->text
- Prompt price
- $2.50 / M tokens
- Entered the catalogue
- 13 May 2024
- Last tested
- 5 October 2026
OpenAI: GPT-4o (2024-08-06)openai/gpt-4o-2024-08-06builds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 128,000
- Max output
- 16,384
- Modalities
- text+image+file->text
- Prompt price
- $2.50 / M tokens
- Entered the catalogue
- 6 August 2024
- Last tested
- 5 October 2026
OpenAI: GPT-4o (2024-11-20)openai/gpt-4o-2024-11-20builds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 128,000
- Max output
- 16,384
- Modalities
- text+image+file->text
- Prompt price
- $2.50 / M tokens
- Entered the catalogue
- 20 November 2024
- Last tested
- 5 October 2026
OpenAI: GPT-4o-miniopenai/gpt-4o-minibuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 128,000
- Max output
- 16,384
- Modalities
- text+image+file->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 18 July 2024
- Last tested
- 5 October 2026
OpenAI: GPT-4o-mini (2024-07-18)openai/gpt-4o-mini-2024-07-18builds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 128,000
- Max output
- 16,384
- Modalities
- text+image+file->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 18 July 2024
- Last tested
- 21 August 2026
OpenAI: GPT-5openai/gpt-5builds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $1.25 / M tokens
- Entered the catalogue
- 7 August 2025
- Last tested
- 5 October 2026
OpenAI: GPT-5 Miniopenai/gpt-5-minibuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 7 August 2025
- Last tested
- 5 October 2026
OpenAI: GPT-5 Nanoopenai/gpt-5-nanobuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 7 August 2025
- Last tested
- 5 October 2026
OpenAI: GPT-5.1openai/gpt-5.1builds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $1.25 / M tokens
- Entered the catalogue
- 13 November 2025
- Last tested
- 5 October 2026
OpenAI: GPT-5.1-Codexopenai/gpt-5.1-codexbuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image->text
- Prompt price
- $1.25 / M tokens
- Entered the catalogue
- 13 November 2025
- Last tested
- 3 September 2026
OpenAI: GPT-5.1-Codex-Maxopenai/gpt-5.1-codex-maxbuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image->text
- Prompt price
- $1.25 / M tokens
- Entered the catalogue
- 4 December 2025
- Last tested
- 3 September 2026
OpenAI: GPT-5.1-Codex-Miniopenai/gpt-5.1-codex-minibuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image->text
- Prompt price
- $0.25 / M tokens
- Entered the catalogue
- 13 November 2025
- Last tested
- 21 August 2026
OpenAI: GPT-5.2openai/gpt-5.2builds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $1.75 / M tokens
- Entered the catalogue
- 10 December 2025
- Last tested
- 5 October 2026
OpenAI: GPT-5.2-Codexopenai/gpt-5.2-codexbuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image->text
- Prompt price
- $1.75 / M tokens
- Entered the catalogue
- 14 January 2026
- Last tested
- 3 September 2026
OpenAI: GPT-5.3-Codexopenai/gpt-5.3-codexbuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $1.75 / M tokens
- Entered the catalogue
- 24 February 2026
- Last tested
- 5 October 2026
OpenAI: GPT-5.4openai/gpt-5.4builds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $2.50 / M tokens
- Entered the catalogue
- 5 March 2026
- Last tested
- 5 October 2026
OpenAI: GPT-5.4 Miniopenai/gpt-5.4-minibuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $0.75 / M tokens
- Entered the catalogue
- 17 March 2026
- Last tested
- 5 October 2026
OpenAI: GPT-5.4 Nanoopenai/gpt-5.4-nanobuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 400,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 17 March 2026
- Last tested
- 21 August 2026
OpenAI: GPT-5.5openai/gpt-5.5builds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $5.00 / M tokens
- Entered the catalogue
- 24 April 2026
- Last tested
- 5 October 2026
OpenAI: GPT-5.5 Proopenai/gpt-5.5-probuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $30.0 / M tokens
- Entered the catalogue
- 24 April 2026
- Last tested
- 5 October 2026
OpenAI: GPT-5.6 Lunaopenai/gpt-5.6-lunabuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 9 July 2026
- Last tested
- 5 October 2026
OpenAI: GPT-5.6 Solopenai/gpt-5.6-solbuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 9 July 2026
- Last tested
- 3 September 2026
OpenAI: GPT-5.6 Terraopenai/gpt-5.6-terrabuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 9 July 2026
- Last tested
- 3 September 2026
OpenAI: GPT-6 Astraopenai/gpt-6-astrabuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $10.0 / M tokens
- Entered the catalogue
- 4 September 2026
- Last tested
- 5 October 2026
OpenAI: GPT-6 Lunaopenai/gpt-6-lunabuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 22 September 2026
- Last tested
- 5 October 2026
OpenAI: GPT-6 Solopenai/gpt-6-solbuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 22 September 2026
- Last tested
- 5 October 2026
OpenAI: GPT-6.1 Solopenai/gpt-6.1-solbuilds on the GPT tokenizerModerate
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with openai/gpt-5-nano, short of a full exact signature
- Context
- 1,050,000
- Max output
- 128,000
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 29 September 2026
- Last tested
- 5 October 2026
OpenAI: gpt-oss-120bopenai/gpt-oss-120bbuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.04 / M tokens
- Entered the catalogue
- 5 August 2025
- Last tested
- 5 October 2026
OpenAI: gpt-oss-20bopenai/gpt-oss-20bbuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 131,072
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.02 / M tokens
- Entered the catalogue
- 5 August 2025
- Last tested
- 21 August 2026
OpenAI: gpt-oss-safeguard-20bopenai/gpt-oss-safeguard-20bbuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 131,072
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 29 October 2025
- Last tested
- 21 August 2026
OpenAI: o3openai/o3builds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 200,000
- Max output
- 100,000
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 16 April 2025
- Last tested
- 5 October 2026
OpenAI: o3 Miniopenai/o3-minibuilds on the GPT tokenizerHigh
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 200,000
- Max output
- 100,000
- Modalities
- text+file->text
- Prompt price
- $1.10 / M tokens
- Entered the catalogue
- 31 January 2025
- Last tested
- 3 September 2026
OpenAI: o3 Mini Highopenai/o3-mini-highbuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 200,000
- Max output
- 100,000
- Modalities
- text+file->text
- Prompt price
- $1.10 / M tokens
- Entered the catalogue
- 12 February 2025
- Last tested
- 3 September 2026
OpenAI: o4 Miniopenai/o4-minibuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 200,000
- Max output
- 100,000
- Modalities
- text+image+file->text
- Prompt price
- $1.10 / M tokens
- Entered the catalogue
- 16 April 2025
- Last tested
- 3 September 2026
OpenAI: o4 Mini Highopenai/o4-mini-highbuilds on the GPT tokenizerVery high
- Maker
- OpenAI
- Label on the catalogue
- GPT
- What the measurement says
- builds on the GPT tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_d1411c9d · shared with 40 other models
- Context
- 200,000
- Max output
- 100,000
- Modalities
- text+image+file->text
- Prompt price
- $1.10 / M tokens
- Entered the catalogue
- 16 April 2025
- Last tested
- 3 September 2026
PrismML: Ternary Bonsai 2 27Bprism-ml/ternary-bonsai-2-27bbuilds on the Qwen tokenizerHigh
- Maker
- Prism Ml
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text+image->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 18 September 2026
- Last tested
- 5 October 2026
Qwen: Qwen Plus 0728qwen/qwen-plus-2025-07-28builds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 1,000,000
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.26 / M tokens
- Entered the catalogue
- 8 September 2025
- Last tested
- 21 August 2026
Qwen: Qwen-Plusqwen/qwen-plusbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 1,000,000
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.26 / M tokens
- Entered the catalogue
- 1 February 2025
- Last tested
- 21 August 2026
Qwen: Qwen2.5 7B Instructqwen/qwen-2.5-7b-instructbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 32,768
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 16 October 2024
- Last tested
- 21 August 2026
Qwen: Qwen2.5 VL 72B Instructqwen/qwen2.5-vl-72b-instructbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 128,000
- Max output
- undeclared
- Modalities
- text+image->text
- Prompt price
- $0.80 / M tokens
- Entered the catalogue
- 1 February 2025
- Last tested
- 3 September 2026
Qwen: Qwen3 14Bqwen/qwen3-14bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.12 / M tokens
- Entered the catalogue
- 28 April 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 235B A22Bqwen/qwen3-235b-a22bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 131,072
- Max output
- 8,192
- Modalities
- text->text
- Prompt price
- $0.45 / M tokens
- Entered the catalogue
- 28 April 2025
- Last tested
- 3 September 2026
Qwen: Qwen3 235B A22B Instruct 2507qwen/qwen3-235b-a22b-2507builds on the Qwen tokenizerVery high
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.09 / M tokens
- Entered the catalogue
- 21 July 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 235B A22B Thinking 2507qwen/qwen3-235b-a22b-thinking-2507builds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 131,072
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.23 / M tokens
- Entered the catalogue
- 25 July 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 30B A3Bqwen/qwen3-30b-a3bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.12 / M tokens
- Entered the catalogue
- 28 April 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 30B A3B Instruct 2507qwen/qwen3-30b-a3b-instruct-2507builds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- 32,000
- Modalities
- text->text
- Prompt price
- $0.05 / M tokens
- Entered the catalogue
- 29 July 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 30B A3B Thinking 2507qwen/qwen3-30b-a3b-thinking-2507builds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 81,920
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 28 August 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 32Bqwen/qwen3-32bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.08 / M tokens
- Entered the catalogue
- 28 April 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 8Bqwen/qwen3-8bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 131,072
- Max output
- 8,192
- Modalities
- text->text
- Prompt price
- $0.12 / M tokens
- Entered the catalogue
- 28 April 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 Coder 30B A3B Instructqwen/qwen3-coder-30b-a3b-instructbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 31 July 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 Coder 480B A35Bqwen/qwen3-coderbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 23 July 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 Coder Flashqwen/qwen3-coder-flashbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 1,000,000
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 17 September 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 Coder Nextqwen/qwen3-coder-nextbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.12 / M tokens
- Entered the catalogue
- 4 February 2026
- Last tested
- 21 August 2026
Qwen: Qwen3 Coder Plusqwen/qwen3-coder-plusbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 1,000,000
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $0.65 / M tokens
- Entered the catalogue
- 23 September 2025
- Last tested
- 5 October 2026
Qwen: Qwen3 Maxqwen/qwen3-maxbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $0.78 / M tokens
- Entered the catalogue
- 23 September 2025
- Last tested
- 3 September 2026
Qwen: Qwen3 Max Thinkingqwen/qwen3-max-thinkingbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $0.78 / M tokens
- Entered the catalogue
- 9 February 2026
- Last tested
- 3 September 2026
Qwen: Qwen3 Next 80B A3B Instructqwen/qwen3-next-80b-a3b-instructbuilds on the Qwen tokenizerVery high
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why very high
- an exact fingerprint shared with 40 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 11 September 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 Next 80B A3B Thinkingqwen/qwen3-next-80b-a3b-thinkingbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 11 September 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 VL 235B A22B Instructqwen/qwen3-vl-235b-a22b-instructbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text+image->text
- Prompt price
- $0.21 / M tokens
- Entered the catalogue
- 23 September 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 VL 235B A22B Thinkingqwen/qwen3-vl-235b-a22b-thinkingbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 131,072
- Max output
- 32,768
- Modalities
- text+image->text
- Prompt price
- $0.40 / M tokens
- Entered the catalogue
- 23 September 2025
- Last tested
- 3 September 2026
Qwen: Qwen3 VL 30B A3B Instructqwen/qwen3-vl-30b-a3b-instructbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- 16,384
- Modalities
- text+image->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 6 October 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 VL 30B A3B Thinkingqwen/qwen3-vl-30b-a3b-thinkingbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text+image->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 6 October 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 VL 32B Instructqwen/qwen3-vl-32b-instructbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 131,072
- Max output
- 32,768
- Modalities
- text+image->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 23 October 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 VL 8B Instructqwen/qwen3-vl-8b-instructbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text+image->text
- Prompt price
- $0.12 / M tokens
- Entered the catalogue
- 14 October 2025
- Last tested
- 21 August 2026
Qwen: Qwen3 VL 8B Thinkingqwen/qwen3-vl-8b-thinkingbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 131,072
- Max output
- 32,768
- Modalities
- text+image->text
- Prompt price
- $0.18 / M tokens
- Entered the catalogue
- 14 October 2025
- Last tested
- 21 August 2026
Qwen: Qwen3.5 397B A17Bqwen/qwen3.5-397b-a17bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image+video->text
- Prompt price
- $0.55 / M tokens
- Entered the catalogue
- 16 February 2026
- Last tested
- 22 August 2026
Qwen: Qwen3.5 Plus 2026-02-15qwen/qwen3.5-plus-02-15builds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 65,536
- Modalities
- text+image+video->text
- Prompt price
- $0.26 / M tokens
- Entered the catalogue
- 16 February 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.5 Plus 2026-04-20qwen/qwen3.5-plus-20260420builds on the Qwen tokenizerVery high
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 65,536
- Modalities
- text+image+video->text
- Prompt price
- $0.30 / M tokens
- Entered the catalogue
- 27 April 2026
- Last tested
- 5 October 2026
Qwen: Qwen3.5-122B-A10Bqwen/qwen3.5-122b-a10bbuilds on the Qwen tokenizerModerate
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why moderate
- near-exact agreement (48 of 48 strings) with nex-agi/nex-n2-mini, short of a full exact signature
- Context
- 262,144
- Max output
- 65,536
- Modalities
- text+image+video->text
- Prompt price
- $0.26 / M tokens
- Entered the catalogue
- 25 February 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.5-27Bqwen/qwen3.5-27bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 262,144
- Max output
- 65,536
- Modalities
- text+image+video->text
- Prompt price
- $0.20 / M tokens
- Entered the catalogue
- 25 February 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.5-35B-A3Bqwen/qwen3.5-35b-a3bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image+video->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 25 February 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.5-9Bqwen/qwen3.5-9bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 262,144
- Max output
- 32,768
- Modalities
- text+image+video->text
- Prompt price
- $0.10 / M tokens
- Entered the catalogue
- 10 March 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.5-Flashqwen/qwen3.5-flash-02-23builds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 65,536
- Modalities
- text+image+video->text
- Prompt price
- $0.07 / M tokens
- Entered the catalogue
- 25 February 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.6 27Bqwen/qwen3.6-27bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 5 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_06ffa14a · shared with 5 other models
- Context
- 262,144
- Max output
- 262,140
- Modalities
- text+image+video->text
- Prompt price
- $0.32 / M tokens
- Entered the catalogue
- 27 April 2026
- Last tested
- 3 September 2026
Qwen: Qwen3.6 35B A3Bqwen/qwen3.6-35b-a3bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 5 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_06ffa14a · shared with 5 other models
- Context
- 262,144
- Max output
- undeclared
- Modalities
- text+image+video->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 27 April 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.6 Flashqwen/qwen3.6-flashbuilds on the Qwen tokenizerVery high
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 65,536
- Modalities
- text+image+video->text
- Prompt price
- $0.19 / M tokens
- Entered the catalogue
- 27 April 2026
- Last tested
- 5 October 2026
Qwen: Qwen3.6 Max Previewqwen/qwen3.6-max-previewbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 262,144
- Max output
- 65,536
- Modalities
- text->text
- Prompt price
- $1.03 / M tokens
- Entered the catalogue
- 27 April 2026
- Last tested
- 5 October 2026
Qwen: Qwen3.6 Plusqwen/qwen3.6-plusbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen3
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 65,536
- Modalities
- text+image+video->text
- Prompt price
- $0.33 / M tokens
- Entered the catalogue
- 2 April 2026
- Last tested
- 5 October 2026
Qwen: Qwen3.7 Flashqwen/qwen3.7-flashbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 65,536
- Modalities
- text+image+video->text
- Prompt price
- $0.03 / M tokens
- Entered the catalogue
- 27 July 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.7 Maxqwen/qwen3.7-maxbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $1.48 / M tokens
- Entered the catalogue
- 21 May 2026
- Last tested
- 5 October 2026
Qwen: Qwen3.7 Plusqwen/qwen3.7-plusbuilds on the Qwen tokenizerVery high
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 131,072
- Modalities
- text+image->text
- Prompt price
- $0.32 / M tokens
- Entered the catalogue
- 3 June 2026
- Last tested
- 21 August 2026
Qwen: Qwen3.8 2.4T A95Bqwen/qwen3.8-2.4t-a95bbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 21 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,048,576
- Max output
- 131,072
- Modalities
- text->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 12 August 2026
- Last tested
- 5 October 2026
Qwen: Qwen3.8 Flashqwen/qwen3.8-flashbuilds on the Qwen tokenizerVery high
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 26 August 2026
- Last tested
- 5 October 2026
Qwen: Qwen3.8 Max (0902)qwen/qwen3.8-max-0902builds on the Qwen tokenizerVery high
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 3 September 2026
- Last tested
- 5 October 2026
Qwen: Qwen3.8 Max Primeqwen/qwen3.8-max-primebuilds on the Qwen tokenizerVery high
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why very high
- an exact fingerprint shared with 21 models and an identical API contract inside the same group: two signals of different kinds agree
- Fingerprint
- tk_181907f2 · shared with 21 other models
- Context
- 1,000,000
- Max output
- 131,072
- Modalities
- text+image+video->text
- Prompt price
- $4.00 / M tokens
- Entered the catalogue
- 23 September 2026
- Last tested
- 5 October 2026
Qwen: Qwen3.8 Omni Flashqwen/qwen3.8-omni-flashbuilds on the Qwen tokenizerModerate
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why moderate
- near-exact agreement (49 of 49 strings) with qwen/qwen3.8-27b, short of a full exact signature
- Context
- 1,000,000
- Max output
- 131,072
- Modalities
- text+image+audio+video->text
- Prompt price
- $0.15 / M tokens
- Entered the catalogue
- 21 September 2026
- Last tested
- 5 October 2026
Qwen2.5 72B Instructqwen/qwen-2.5-72b-instructbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 32,768
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.36 / M tokens
- Entered the catalogue
- 19 September 2024
- Last tested
- 21 August 2026
Qwen2.5 Coder 32B Instructqwen/qwen-2.5-coder-32b-instructbuilds on the Qwen tokenizerHigh
- Maker
- Alibaba (Qwen)
- Label on the catalogue
- Qwen
- What the measurement says
- builds on the Qwen tokenizer
- Why high
- an exact fingerprint shared with 40 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5410b38e · shared with 40 other models
- Context
- 32,768
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.66 / M tokens
- Entered the catalogue
- 11 November 2024
- Last tested
- 3 September 2026
ReMM SLERP 13Bundi95/remm-slerp-l2-13bbuilds on the Llama2 tokenizerHigh
- Maker
- Undi95
- Label on the catalogue
- Llama2
- What the measurement says
- builds on the Llama2 tokenizer
- Why high
- an exact fingerprint shared with 1 model that declares the same family, without a second signal of a different kind yet
- Fingerprint
- tk_5348f9f6 · shared with 1 other model
- Context
- 6,144
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.35 / M tokens
- Entered the catalogue
- 22 July 2023
- Last tested
- 22 August 2026
Sao10K: Llama 3 8B Lunarissao10k/l3-lunaris-8bbuilds on the Llama3 tokenizerHigh
- Maker
- Sao10k
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- one template-boundary shift from a group whose members all declare the same family
- Fingerprint
- tk_066351de · unique so far
- Context
- 8,192
- Max output
- undeclared
- Modalities
- text->text
- Prompt price
- $0.04 / M tokens
- Entered the catalogue
- 13 August 2024
- Last tested
- 21 August 2026
Sao10K: Llama 3.1 Euryale 70B v2.2sao10k/l3.1-euryale-70bbuilds on the Llama3 tokenizerHigh
- Maker
- Sao10k
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 8 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_6588ed5b · shared with 8 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.85 / M tokens
- Entered the catalogue
- 28 August 2024
- Last tested
- 21 August 2026
Sao10K: Llama 3.3 Euryale 70Bsao10k/l3.3-euryale-70bbuilds on the Llama3 tokenizerHigh
- Maker
- Sao10k
- Label on the catalogue
- Llama3
- What the measurement says
- builds on the Llama3 tokenizer
- Why high
- an exact fingerprint shared with 8 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_6588ed5b · shared with 8 other models
- Context
- 131,072
- Max output
- 16,384
- Modalities
- text->text
- Prompt price
- $0.65 / M tokens
- Entered the catalogue
- 18 December 2024
- Last tested
- 21 August 2026
SpaceXAI: Grok 4.20x-ai/grok-4.20builds on the Grok tokenizerHigh
- Maker
- xAI
- Label on the catalogue
- Grok
- What the measurement says
- builds on the Grok tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_a17bbc1d · shared with 2 other models
- Context
- 2,000,000
- Max output
- undeclared
- Modalities
- text+image+file->text
- Prompt price
- $1.25 / M tokens
- Entered the catalogue
- 31 March 2026
- Last tested
- 3 September 2026
SpaceXAI: Grok 4.3x-ai/grok-4.3builds on the Grok tokenizerHigh
- Maker
- xAI
- Label on the catalogue
- Grok
- What the measurement says
- builds on the Grok tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_a17bbc1d · shared with 2 other models
- Context
- 1,000,000
- Max output
- undeclared
- Modalities
- text+image+file->text
- Prompt price
- $1.25 / M tokens
- Entered the catalogue
- 30 April 2026
- Last tested
- 5 October 2026
SpaceXAI: Grok 4.5x-ai/grok-4.5builds on the Grok tokenizerHigh
- Maker
- xAI
- Label on the catalogue
- Grok
- What the measurement says
- builds on the Grok tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_cb7f8e86 · shared with 2 other models
- Context
- 500,000
- Max output
- undeclared
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 8 July 2026
- Last tested
- 5 October 2026
SpaceXAI: Grok 4.6x-ai/grok-4.6builds on the Grok tokenizerHigh
- Maker
- xAI
- Label on the catalogue
- Grok
- What the measurement says
- builds on the Grok tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_cb7f8e86 · shared with 2 other models
- Context
- 500,000
- Max output
- undeclared
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 12 August 2026
- Last tested
- 5 October 2026
SpaceXAI: Grok 4.7x-ai/grok-4.7builds on the Grok tokenizerHigh
- Maker
- xAI
- Label on the catalogue
- Grok
- What the measurement says
- builds on the Grok tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_cb7f8e86 · shared with 2 other models
- Context
- 500,000
- Max output
- undeclared
- Modalities
- text+image+file->text
- Prompt price
- $2.00 / M tokens
- Entered the catalogue
- 21 September 2026
- Last tested
- 5 October 2026
SpaceXAI: Grok Build 0.1x-ai/grok-build-0.1builds on the Grok tokenizerHigh
- Maker
- xAI
- Label on the catalogue
- Grok
- What the measurement says
- builds on the Grok tokenizer
- Why high
- an exact fingerprint shared with 2 models that declare the same family, without a second signal of a different kind yet
- Fingerprint
- tk_a17bbc1d · shared with 2 other models
- Context
- 256,000
- Max output
- undeclared
- Modalities
- text+image+file->text
- Prompt price
- $1.00 / M tokens
- Entered the catalogue
- 20 May 2026
- Last tested
- 3 September 2026
Nothing matches that. Clear the box to see the whole catalogue.
Why the dull ones are here too
Plenty of these rows say the obvious: a maker's model reads with that maker's own vocabulary, exactly as its label claims. Those are published because a measurement we took is a measurement we show, and because they are the reference library that makes the interesting ones possible. Ox Alpha could only be placed with the Z.ai family because GLM had already been measured. The method and the controls cover both halves.
Using these figures
Quote any figure here with attribution to YFarmX and a link to the row it came from. Every row has its own address, so a cited number stays checkable, and measurements are dated and append-only. We are not handing out the raw files at the moment; for bulk or commercial use, ask first.