Anthropic: Claude Sonnet 5

anthropic/claude-sonnet-5 · AnthropicNot yet established

Tokenizer lineage (measured): Claude
Evidence confidence: Moderate
Declared tokenizer tag: Claude
Entered catalogue: 30 June 2026
Last tested: 3 September 2026
Measurement: Measured

Evidence

✓ near-exact agreement (49/49) with anthropic/claude-opus-4.8

Compare with Anthropic: Claude Opus 4.8

Download JSON Report a correction

Identity statement

Anthropic: Claude Sonnet 5 is Anthropic's model. Its measured tokenizer sits with the Claude group, which is evidence that it builds on the Claude tokenizer. Tokenizer reuse is normal engineering, open vocabularies travel between labs, and this says nothing against Anthropic's authorship of the model itself.

Evidence stack

Signals of different kinds, weighed together

Signal	Result	Reference
Tokenizer signature	Measured with 1 corrupt row excluded	–
API surface	Identical contract to 2 other entries	api_5d3cdea0
Context and output	1,000,000 context · 128,000 max output	declared
Reasoning contract	Optional · efforts max, xhigh, high, medium, low · default high	declared
Serving providers observed	Claude Platform on AWS	measured 3 September 2026

Closest measured models

Agreement across the mutually clean test strings

Anthropic: Claude Opus 4.8anthropic/claude-opus-4.849/49 · 100%Anthropic: Claude Fable 5anthropic/claude-fable-544/44 · 100%Anthropic: Claude Opus 5anthropic/claude-opus-547/49 · 96%Anthropic: Claude Fable 5.1anthropic/claude-fable-5.147/49 · 96%Anthropic: Claude Opus 4.7anthropic/claude-opus-4.727/49 · 55%IBM: Granite 4.0 Microibm-granite/granite-4.0-h-micro15/49 · 31%

Click a row to open the full comparison.

The measured fingerprint

Marginal prompt-token cost of each test string, grouped by script

All 50 rows, 1 excluded as corrupt

English and whitespace

en-prose24

en-long28

spaces-204

spaces-606

tabs-204

newlines-204

mixed-ws9

Digits

digits-93

digits-124

digits-3010

digits-sep7

float-long9

CJK

zh-common18

zh-long24

zh-rare20

ja-kana12

ja-kanji16

ko19

Other scripts

ru14

ar21

he18

hi23

th25

el22

Emoji

emoji-basic10

emoji-skin25

emoji-zwj-family14

emoji-zwj-x342

emoji-flags24

emoji-prof27

Rare Unicode

math46

boxdraw28

combining6

cjk-ext-b16

surrogates54

zalgo36

rtl-mix11

Code

py-code30

py-indent19

json29

html18

regex63

camel16

snake14

Repetition and encodings

rare-word-x556

repeat-tok61

base64x

hex18

url26

uuid27

An x marks a row excluded as corrupt (caching or provider interference detected during measurement). Overhead subtracted: 6 prompt tokens.

Declared record

What the catalogue claims about this model

Sonnet 5 is Anthropic's most capable Sonnet-class model, with frontier performance across coding, agents, and professional work. It supports adaptive thinking with selectable reasoning effort levels (low, medium, high, max,...

Modalities: text+image+file->text
Declared tokenizer: Claude
Prompt price: $2.00 / M tokens
Completion price: $10.0 / M tokens

Supported parameters

include_reasoningmax_completion_tokensmax_tokensreasoningreasoning_effortresponse_formatstopstructured_outputstool_choicetoolsverbosity

defaults: {}

Catalogue entry

History

Every observation, kept as taken

30 June 2026Enters the OpenRouter catalogue declared as Claude.
3 September 2026Fingerprinted in the YFarmX catalogue sweep · 49 of 50 strings measured clean.
3 September 2026YFarmX assessment: consistent with the Claude family, moderate confidence.

Cite this page as the evidence record for Anthropic: Claude Sonnet 5: the URL is stable, measurements are dated, and revisions append to the history above. Method: tokenizer fingerprinting, v1, 50 strings.