DeepSeek: R1

deepseek/deepseek-r1 · DeepSeekNot yet established

Tokenizer group (measured): DeepSeek
Declared tokenizer tag: DeepSeek
Entered catalogue: 20 January 2025
Last tested: 5 October 2026
Evidence confidence: Moderate

Why moderate: near-exact agreement (47 of 50 strings) with deepseek/deepseek-r1-distill-llama-70b, short of a full exact signature.

Evidence

✓ near-exact agreement (47/50) with deepseek/deepseek-r1-distill-llama-70b

Compare with DeepSeek: R1 Distill Llama 70B

Report a correction

Identity statement

DeepSeek: R1 is DeepSeek's model, and its measured tokenizer sits in DeepSeek's own signature group alongside its sibling models. No other maker's measured model shares it so far.

What we checked

Four different kinds of evidence, and what each one showed

We checked	What we found	Where it came from
How it counts tokens	Counted cleanly, and no other model in the catalogue counts the same way	tk_d6c0b2be
How its API is set up	A combination of settings no other listing in the catalogue uses	api_2283b8cd
How much it can read and write	Reads up to 64,000 tokens at once, writes up to 16,000	its own listing
Whether it thinks before answering	Always reasons, and cannot be turned off	its own listing
Who served our requests	Novita	we measured it, 5 October 2026

Closest measured models

Agreement across the mutually clean test strings

DeepSeek: R1 Distill Llama 70Bdeepseek/deepseek-r1-distill-llama-70b47/50 · 94%DeepSeek: DeepSeek V3.1deepseek/deepseek-chat-v3.147/50 · 94%DeepSeek: DeepSeek V3deepseek/deepseek-chat47/50 · 94%DeepSeek: DeepSeek V4 Pro 0813deepseek/deepseek-v4-pro-081345/50 · 90%DeepSeek: DeepSeek V4 Flash Vision Expdeepseek/deepseek-v4-flash-vision-exp45/50 · 90%DeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-073145/50 · 90%

Click a row to open the full comparison.

The measured fingerprint

What each test string cost this model, in tokens

We sent DeepSeek: R1 fifty short pieces of text and recorded what each one cost it in tokens. The bars below are those costs. Two models built on the same tokenizer produce the same bars; a model built on a different one produces a different set, which is what makes this a fingerprint.

All 50 rows

English and whitespace

en-prose14

en-long16

spaces-203

spaces-603

tabs-204

newlines-204

mixed-ws7

Digits

digits-93

digits-124

digits-3010

digits-sep7

float-long9

CJK

zh-common9

zh-long10

zh-rare18

ja-kana9

ja-kanji11

ko13

Other scripts

ru13

ar10

he14

hi18

th11

el18

Emoji

emoji-basic10

emoji-skin14

emoji-zwj-family11

emoji-zwj-x335

emoji-flags16

emoji-prof21

Rare Unicode

math22

boxdraw27

combining10

cjk-ext-b16

surrogates33

zalgo36

rtl-mix6

Code

py-code25

py-indent15

json20

html16

regex47

camel7

snake8

Repetition and encodings

rare-word-x525

repeat-tok40

base6434

hex12

url17

uuid27

Overhead subtracted: 3 prompt tokens.

Declared record

What the catalogue claims about this model

DeepSeek R1 is here: Performance on par with [OpenAI o1](/openai/o1), but open-sourced and with fully open reasoning tokens. It's 671B parameters in size, with 37B active in an inference pass....

Modalities: text->text
Declared tokenizer: DeepSeek
Prompt price: $0.70 / M tokens
Completion price: $2.50 / M tokens

Supported parameters

frequency_penaltyinclude_reasoningmax_tokenspresence_penaltyreasoningrepetition_penaltyresponse_formatseedstoptemperaturetool_choicetoolstop_ktop_p

defaults: {"temperature":null,"top_p":null,"top_k":null,"frequency_penalty":null,"presence_penalty":null,"repetition_penalty":null}

Catalogue entry Weights on Hugging Face

History

Every observation, kept as taken

20 January 2025Enters the OpenRouter catalogue declared as DeepSeek.
5 October 2026Fingerprinted in the YFarmX catalogue sweep · 50 of 50 strings measured clean.
5 October 2026YFarmX assessment: consistent with the DeepSeek family, moderate confidence.