OpenAI: o3

openai/o3 · OpenAIDeclared and measured agree

Tokenizer lineage (measured): GPT
Declared tokenizer tag: GPT
Entered catalogue: 16 April 2025
Last tested: 5 October 2026
Evidence confidence: Very high

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.

Evidence

✓ exact tokenizer signature shared with 40 models
✓ API-surface contract identical to a group member

Compare with OpenAI: GPT-6 Luna

Report a correction

Identity statement

OpenAI: o3 is OpenAI's model. Its measured tokenizer sits with the GPT group, which is evidence that it builds on the GPT tokenizer. Tokenizer reuse is normal engineering, open vocabularies travel between labs, and this says nothing against OpenAI's authorship of the model itself.

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	Counts every one of the 50 test strings exactly like 40 other models	tk_d1411c9d
How its API is set up	Set up identically to 3 other listings, field for field	api_a2a7f2ca
How much it can read and write	Reads up to 200,000 tokens at once, writes up to 100,000	its own listing
Whether it thinks before answering	Reasons when asked to	its own listing
Who served our requests	OpenAI	we measured it, 5 October 2026

Shares this fingerprint

Exact signature first; template-boundary shifts of the same signature beneath

Closest measured models

Agreement across the mutually clean test strings

OpenAI: GPT-5 Nanoopenai/gpt-5-nano50/50 · 100%OpenAI: GPT-5.5 Proopenai/gpt-5.5-pro50/50 · 100%OpenAI: gpt-oss-20bopenai/gpt-oss-20b50/50 · 100%OpenAI: gpt-oss-120bopenai/gpt-oss-120b50/50 · 100%OpenAI: gpt-oss-safeguard-20bopenai/gpt-oss-safeguard-20b50/50 · 100%OpenAI: GPT-4.1 Nanoopenai/gpt-4.1-nano50/50 · 100%

Click a row to open the full comparison.

The measured fingerprint

What each test string cost this model, in tokens

We sent OpenAI: o3 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-long17

spaces-203

spaces-603

tabs-203

newlines-204

mixed-ws6

Digits

digits-93

digits-124

digits-3010

digits-sep7

float-long9

CJK

zh-common12

zh-long17

zh-rare20

ja-kana10

ja-kanji13

ko11

Other scripts

ru11

ar11

he10

hi11

th11

el14

Emoji

emoji-basic8

emoji-skin13

emoji-zwj-family11

emoji-zwj-x333

emoji-flags16

emoji-prof20

Rare Unicode

math31

boxdraw27

combining11

cjk-ext-b13

surrogates40

zalgo35

rtl-mix5

Code

py-code25

py-indent15

json20

html16

regex44

camel6

snake6

Repetition and encodings

rare-word-x531

repeat-tok22

base6434

hex11

url17

uuid27

Overhead subtracted: 6 prompt tokens.

Declared record

What the catalogue claims about this model

o3 is a well-rounded and powerful model across domains. It sets a new standard for math, science, coding, and visual reasoning tasks. It also excels at technical writing and instruction-following....

Modalities: text+image+file->text
Declared tokenizer: GPT
Prompt price: $2.00 / M tokens
Completion price: $8.00 / M tokens

Supported parameters

include_reasoningmax_tokensreasoningresponse_formatseedstructured_outputstool_choicetools

defaults: {}

Catalogue entry

History

Every observation, kept as taken

16 April 2025Enters the OpenRouter catalogue declared as GPT.
5 October 2026Fingerprinted in the YFarmX catalogue sweep · 50 of 50 strings measured clean.
5 October 2026YFarmX assessment: consistent with the GPT family, very high confidence.