Pass IndexThe State of AISign in

o3

OpenAI's o3 is their most powerful reasoning model, setting new state-of-the-art benchmarks in coding, math, science, and visual perception.

text + image + file → text · made by OpenAI

$2→$8per Mtok in / out
OpenAI · 8 sellers

Sold by 12 ways

SellerLaneRate
Microsoft Azure AI cloudbatch$1per Mtok in$4per Mtok outprices.azure.com · read 2026-08-24
DigitalOcean cloudstandard$2per Mtok in$8per Mtok out$0.5per Mtok cachedmodels.dev · read 2026-09-16
Microsoft Azure AI cloudstandard$2per Mtok in$8per Mtok out$0.5per Mtok cachedmodels.dev · read 2026-09-16
OpenAI apistandard$2per Mtok in$8per Mtok out$0.5per Mtok cacheddevelopers.openai.com · read 2026-08-24
Requesty aggregatornot on the seller's list todayopenai flex$0.9per Mtok in$3.6per Mtok out$0.23per Mtok cachedrouter.requesty.ai · read 2026-08-25
Requesty aggregatoropenai$2per Mtok in$8per Mtok out$0.5per Mtok cachedrouter.requesty.ai · read 2026-09-10
Nous Research 2 ways$2per Mtok in$8per Mtok out$0.5per Mtok cached
Nous Research aggregatorbatch$1per Mtok in$4per Mtok out$0.25per Mtok cachedinference-api.nousresearch.com · read 2026-09-12
Nous Research aggregatorstandard$2per Mtok in$8per Mtok out$0.5per Mtok cachedinference-api.nousresearch.com · read 2026-09-12
OpenRouter 2 ways$2per Mtok in$8per Mtok out$0.5per Mtok cached
OpenRouter aggregatorbatch$1per Mtok in$4per Mtok out$0.25per Mtok cachedopenrouter.ai · read 2026-08-24
OpenRouter aggregatorstandard$2per Mtok in$8per Mtok out$0.5per Mtok cachedopenrouter.ai · read 2026-08-24
ElectronHub aggregatorstandard$1.5per Mtok in$6per Mtok outapi.electronhub.ai · read 2026-09-16
Vercel AI Gateway aggregatorstandard$2per Mtok in$8per Mtok out$0.5per Mtok cachedai-gateway.vercel.sh · read 2026-08-25

Measured 52 standings

PlaceBoardMetricScore
1stof 15Cad eval — Epoch AIOverall pass (%) (medium)0.74
1stof 42Fictionlivebench — Epoch AI120k token score (medium)1
1stof 81Forecastbench — Epoch AIOverall score62.5
1stof 22SWE-bench Multimodal% Resolved35.98
2ndof 28LiveCodeBench Leaderboard (default window 8/1/2024–5/1/2025, 454 problems)Pass@1 (high)75.8
5thof 108Math level 5 — Epoch AImean_score (high)97.772
6thof 69Aider polyglot coding leaderboardPercent correct (high)81.3
6thof 72Aider polyglot — Epoch AIPercent correct (high)81.3
6thof 49Lech mazur writing — Epoch AIMean score (medium)8.39
7thof 11Gdpval — Epoch AIWin Rate (%) (medium)0.308
8thof 109Berkeley Function-Calling Leaderboard (BFCL) V4Overall Acc (prompt)63.05
10thof 46Enigma eval — Epoch AIAccuracy (medium)13.09
10thof 38Vpct — Epoch AICorrect (medium)0.52
14thof 23Cl bench — Epoch AIOverall (high)0.178
16thof 51Humanity's Last Exam (Epoch AI replication)Accuracy (high)20.32
16thof 20Os world — Epoch AIScore (medium)23
17thof 50Metr time horizons — Epoch AIaverage_score (medium)0.654
21stof 61AIME 2025Accuracy (± 95% CI) (high)89.17
21stof 38Gso — Epoch AIScore OPT@1 (high)0.088
22ndof 80Simpleqa verified — Epoch AImean_score (high)49.4
26thof 34LMArena · SearchScore (Elo)1144
27thof 127Mystery game puzzles — Epoch AImean_score (high)29
30thof 41Deepresearchbench — Epoch AIAverage score (medium)0.452
30thof 35Swe bench verified — Epoch AImean_score (medium)62.319
31stof 222Chess puzzles — Epoch AImean_score (medium)38
36thof 123Lmca — Epoch AIScore (high)39.7
43rdof 65Apex agents — Epoch AIPass@1 score (high)0.172
45thof 101FrontierMath (Tiers 1-3)mean_score (high)18.685
46thof 161Dtbench — Epoch AIAccuracy (high)84.8
47thof 112Ale bench — Epoch AIPerformance (high)933.55
47thof 102Simplebench — Epoch AIScore (AVG@5) (high)0.531
58thof 72Frontiermath tier 4 — Epoch AImean_score (high)2.083
58thof 106Frontiermath tiers 1 3 v2 — Epoch AImean_score (high)33.333
64thof 170Weirdml — Epoch AIAccuracy (high)52.42
68thof 152LMArena · VisionScore (Elo)1215
74thof 152Design Arena · dataviz (models)Elo1200
87thof 290OTIS Mock AIME 2024-2025mean_score (medium)84.444
99thof 402LMArena · TextScore (Elo)1432
100thof 230ARC-AGI-1 (Semi-Private)Score (%) (low)75.7
108thof 177Critpt — Epoch AIAccuracy (high)1.4
111thof 205ARC-AGI-1 (Public Eval)Score (%) (high)64.25
114thof 313GPQA Diamondmean_score (high)81.818
119thof 302Artificial Analysis · Intelligence IndexArtificial Analysis Intelligence Index20
128thof 216ARC-AGI-2 (Public Eval)Score (%) (medium)4.49
128thof 232ARC-AGI-2 (Semi-Private)Score (%) (high)6.53
128thof 244Arc agi — Epoch AIScore (high)60.83
129thof 150Design Arena · uicomponent (models)Elo1032
130thof 227Arc agi 2 — Epoch AIScore (high)6.53
130thof 155Design Arena · gamedev (models)Elo1057
136thof 156Design Arena · codecategories (models)Elo1035
138thof 162Design Arena · website (models)Elo1049
267thof 656Epoch capabilities index — Epoch AIECI Score (high)146.91

About

o3 — a text model from OpenAI, sold by 8 companies from $1.5 in and $6 out per million tokens, placed 1st of 81 on Forecastbench — Epoch AI.

It takes text, images and files and returns text, with a context window of 200,000 tokens. It was published in June 2024, trained on material up to May 2024. Its sellers say it can reason step by step and call a tool. The catalogue files it under reasoning, code and agents. Eight companies sell it. The cheapest is $1.5 in and $6 out per million tokens at ElectronHub. Beside the standard rate there are batch and separately routed lanes. It has been measured on 52 boards, and stands best at 1st of 81 on Forecastbench — Epoch AI.

Every current figure

Maker
OpenAI
Register
model
Takes
text + image + file
Returns
text
Context
200,000 tokens
Longest answer
131,072 tokens
Published
June 2024
Knowledge to
May 2024
Licence
not read
Sellers
8
Price
$1.5 in and $6 out per million tokens — ElectronHub
Boards
52
Best place
1st of 81 — Forecastbench — Epoch AI

Known as 10 names

o3 (high)o3-2025-04-16 (Prompt)o3 (medium)o3-searchGUIRepair + o3 (2025-04-16)o3 0416openai/o3openai/o3:batchopenai/o3:flexOpenAI: o3 (batch)