Models / o3

OpenAI

o3

o3 is best for agents and tool use; pick it when quality matters more than cost. The cheapest model of similar intelligence is Gemma 4 26B A4B, at $0.14 per million tokens against $3.50.

Retiring 11 Dec 2026. OpenAI retires o3; the official replacement is gpt-5.6-sol. How to re-test before you switch.

Results 16 Nov 2026 · 43 results · 5 sources

Price per million tokens
$2.00 in · $8.00 out
Typical host on OpenRouter, 10 Nov 2026 · compare prices
Speed
155 tokens a second; on its test prompt the answer starts after 4.6 s, thinking time included
Artificial Analysis, on its usual API
Intelligence Index
20.2
Artificial Analysis
Spring Prompt overall
Not ranked yet
needs our benchmarks and two groups of results
Developer
OpenAI, based in the US
Weights not published
Released
16 Apr 2025
200,000 tokens of context

Where it stands among the models you could choose

Available models grouped by price, intelligence, speed and developer location. Retired models are excluded.

AmongIntelligence IndexSpring Prompt overallPrice, blendedSpeed
Models available today 100th of 185Claude Opus 5.5 – $3.50 blended, 82nd percentilecheapest: Mistral Nemo 20th of 65Nemotron 3.5 Super VL
Models priced $3–10 per million tokens 30th of 38Claude Opus 5.5 – sets this group 2nd of 15Mistral Medium 3.5
Models with similar intelligence sets this group – $3.50 blended, 90th percentilecheapest: Gemma 4 26B A4B 2nd of 6Gemini 3.5 Flash-Lite
Models writing 100–200 tokens a second 14th of 21Claude Sonnet 5.5 – $3.50 blended, 75th percentilecheapest: Qwen3.5-9B sets this group
Models from developers based in the US 57th of 92Claude Opus 5.5 – $3.50 blended, 66th percentilecheapest: Gemma 3 4B 12th of 33Nemotron 3.5 Super VL

The name under each rank is the leader of that group; green is the top quarter of the group and red the bottom quarter, by rank. Price is shown as a percentile: the share of the group that costs less, so lower is cheaper. "Sets this group" marks the measure the group is defined by. Spring Prompt overall is our score out of 100 across every source (how it works). Price is the developer's list price (or the typical OpenRouter host where we haven't read one) for three input tokens to one output; intelligence and speed are from Artificial Analysis (data sourced from Artificial Analysis); "similar intelligence" means within 4 points on its Intelligence Index. Developer locations are where each company is based, not where a model is served. Groups under 3 models are not ranked.

How good is it, and for what?

Each use case ranks the available models its benchmarks measured, then those priced $3–10 per million tokens. The bar shows its percentile in that field, best to the right. Open a row for the results behind it.

Product listingsTurning a sparse product feed and photos into listings that can go live Not measured
  • CatalogBench: has not measured this model.
Decks from an analysisTurning a finished analysis into a deck you could present as it is Not measured
  • DeckBench: has not measured this model.
User surveysPlanning a user survey and reading its results without being misled Not measured
  • SurveyBench: has not measured this model.
Marketing planningPlanning a year of ad spend without overspending Not measured
  • ROASBench: has not measured this model.
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 16th of 159 available · 11th of 35 at its price
Benchmarko3Rank among availableBest available
Berkeley Function Calling Leaderboard (BFCL) V4BFCL: overall accuracy · reported 63.1% 5th of 36 Claude Opus 4.5 77.5%
  • tau2-bench: has not measured this model.
  • OpenHands Index: has not measured this model.
  • Microsoft STATE-Bench: has not measured this model.
  • Artificial Analysis: has not measured this model.
  • Vending-Bench 2: has not measured this model.
Professional workReal tasks from banking, consulting and law, business documents and freelance projects Not measured
  • APEX-Agents: has not measured this model.
  • GDP.pdf: has not measured this model.
  • Remote Labor Index: has not measured this model.
Reasoning and knowledgeHard questions across science, maths and general knowledge 105th of 187 available · 32nd of 39 at its price
Benchmarko3Rank among availableBest available
Artificial AnalysisArtificial Analysis: Artificial Analysis Intelligence Index · reported 20.2 100th of 185 Claude Opus 5.5 57.6
Artificial AnalysisArtificial Analysis: Humanity's Last Exam · reported 20.1% 103rd of 186 Claude Opus 5.5 61.4%
Artificial AnalysisArtificial Analysis: GPQA Diamond · reported 82.7% 84th of 168 GPT-6 Astra 96.3%
WritingWhat people prefer in blind comparisons, and judged writing quality 52nd of 169 available · 22nd of 34 at its price
Benchmarko3Rank among availableBest available
Arena (formerly LMArena)Arena Text: overall · reported 1,432 73rd of 153 Gemini 4 Argon 1,525
UGI LeaderboardUGI: writing score · reported 66.8 20th of 124 Gemini 3.8 Flash 78.6
Sticking to the factsSummarising without inventing things, and factual answers 88th of 139 available · 25th of 31 at its price
Benchmarko3Rank among availableBest available
Arena (formerly LMArena)Arena Text factuality: overall · reported 1,405 95th of 109 Gemini 4 Argon 1,511
SimpleQA Verified (Epoch AI)SimpleQA Verified: correct answers · reported 49.4% 24th of 68 GPT-6 Astra 75.6%
  • Vectara Hallucination Leaderboard: has not measured this model.
Speed under pressureGood decisions against a real clock (fast chess) Not measured
  • BulletBench: has not measured this model.

Nearest alternatives

Against o3 at $3.50 blended, Intelligence Index 20.2. "Similar intelligence" means within 4 points or better.

Where to run it

1 host on OpenRouter at the standard service tier, with a median of $3.50 blended per million tokens. Cheapest first.

HostQuantisationInputOutputBlended
OpenAInot stated$2.00$8.00$3.50

Prices per million tokens from OpenRouter's public endpoint list, checked 10 Nov 2026.

Against the alternatives

The models you would most likely weigh it against: the leaders of the groups above.

Scroll sideways to see every alternative.

o3Claude Opus 5.5
most intelligent at $3–10 per million tokens
Gemma 4 26B A4B
cheapest with similar intelligence
Gemini 3.5 Flash-Lite
fastest with similar intelligence
Price per million tokens $3.50 $8.00$0.14$0.85
Tokens a second 155 9682398
Intelligence Index 20.2 57.616.722.2
Spring Prompt overall – 82–25
Rank among available models, by use case
Product listings – 13th–25th
Decks from an analysis – 7th–29th
User surveys – 4th–19th
Marketing planning – 9th–28th
Agents and tool use 16th of 159 3rd105th81st
Professional work – 3rd–46th
Reasoning and knowledge 105th of 187 1st110th98th
Writing 52nd of 169 5th82nd49th
Sticking to the facts 88th of 139 3rd43rd61st
Speed under pressure – ––2nd

Shaded figures are better than o3's. A dash means no result.

Where it has been measured

2nd fastest of the 15 models priced $3–10 per million tokens. 30th most intelligent of the 38 models priced $3–10 per million tokens. Among models available today, in the top quarter (by percentile) for agents and tool use.

Every result

43 published results from 5 sources, each in the source's own units, with the configuration that produced it.

Arena (formerly LMArena) · reported by Arena (formerly LMArena) · 8 results
MeasureValueRankConfigurationDated
Arena Search: overallArena rating, higher is better 1,192
1,185–1,199
18 of 34 o3 (arena search) 24 Aug 2026
Arena Text factuality: overallArena rating, higher is better 1,405
1,398–1,411
150 of 182 o3 2 Oct 2026
Arena Text: business, management and financeArena rating, higher is better 1,427
1,420–1,433
114 of 406 o3 2 Oct 2026
Arena Text: creative writingArena rating, higher is better 1,383
1,376–1,390
138 of 411 o3 2 Oct 2026
Arena Text: expert promptsArena rating, higher is better 1,449
1,437–1,460
117 of 363 o3 2 Oct 2026
Arena Text: instruction followingArena rating, higher is better 1,403
1,397–1,408
139 of 413 o3 2 Oct 2026
Arena Text: overallArena rating, higher is better 1,432
1,428–1,435
110 of 413 o3 2 Oct 2026
Arena Text: writing, literature and languageArena rating, higher is better 1,396
1,390–1,402
135 of 412 o3 2 Oct 2026

Not shown: Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.

Artificial Analysis · reported by Artificial Analysis · 9 results
MeasureValueRankConfigurationDated
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 20.2 210 of 413 o3 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 82.7% 153 of 360 o3 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 20.1% 200 of 419 o3 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 71.4% 51 of 288 o3 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 74.7% 154 of 407 o3 20 Oct 2026
Artificial Analysis: Output speedtokens per second, higher is better 155 35 of 137 o3 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 37.1% 67 of 282 o3 20 Oct 2026
Artificial Analysis: Time to first answer tokenseconds, lower is better 4.60 41 of 137 o3 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 80.7% 100 of 286 o3 20 Oct 2026

Not shown: Not business work, and a blend: read the parts for any one task.

Berkeley Function Calling Leaderboard (BFCL) V4 · reported by Berkeley Function Calling Leaderboard (BFCL) V4 · 16 results
MeasureValueRankConfigurationDated
BFCL: irrelevance detection% correct, higher is better 86.1% 28 of 109 o3 (bfcl fc) 16 Dec 2025
BFCL: irrelevance detection% correct, higher is better 84.0% 39 of 109 o3 (bfcl prompt) 16 Dec 2025
BFCL: memory% correct, higher is better 47.3% 12 of 109 o3 (bfcl fc) 16 Dec 2025
BFCL: memory% correct, higher is better 51.8% 10 of 109 o3 (bfcl prompt) 16 Dec 2025
BFCL: multi-turn tasks% correct, higher is better 14.8% 58 of 109 o3 (bfcl fc) 16 Dec 2025
BFCL: multi-turn tasks% correct, higher is better 62.3% 8 of 109 o3 (bfcl prompt) 16 Dec 2025
BFCL: overall accuracy% correct, higher is better 48.6% 30 of 109 o3 (bfcl fc) 16 Dec 2025
BFCL: overall accuracy% correct, higher is better 63.1% 8 of 109 o3 (bfcl prompt) 16 Dec 2025
BFCL: relevance detection% correct, higher is better 81.3% 41 of 109 o3 (bfcl fc) 16 Dec 2025
BFCL: relevance detection% correct, higher is better 93.8% 8 of 109 o3 (bfcl prompt) 16 Dec 2025
BFCL: single-turn calls (curated)% correct, higher is better 40.4% 100 of 109 o3 (bfcl fc) 16 Dec 2025
BFCL: single-turn calls (curated)% correct, higher is better 81.9% 66 of 109 o3 (bfcl prompt) 16 Dec 2025
BFCL: single-turn calls (user-contributed)% correct, higher is better 66.2% 76 of 109 o3 (bfcl fc) 16 Dec 2025
BFCL: single-turn calls (user-contributed)% correct, higher is better 73.2% 50 of 109 o3 (bfcl prompt) 16 Dec 2025
BFCL: web search% correct, higher is better 77.0% 9 of 109 o3 (bfcl fc) 16 Dec 2025
BFCL: web search% correct, higher is better 50.5% 26 of 109 o3 (bfcl prompt) 16 Dec 2025

Not shown: Not general refusal or safety behaviour.

SimpleQA Verified (Epoch AI) · reported by SimpleQA Verified (Epoch AI) · 1 result
MeasureValueRankConfigurationDated
SimpleQA Verified: correct answers% of questions, higher is better 49.4% 25 of 85 o3 (high reasoning) 27 Aug 2026

Not shown: Not answers grounded in your documents; tests what the model remembers.

UGI Leaderboard · reported by UGI Leaderboard · 9 results
MeasureValueRankConfigurationDated
UGI: requested-length error% off the requested word count, lower is better 18.0% 158 of 379 o3 (high reasoning) 24 Sep 2025
UGI: requested-length error% off the requested word count, lower is better 29.0% 257 of 379 o3 (low reasoning) 24 Sep 2025
UGI: requested-length error% off the requested word count, lower is better 18.0% 158 of 379 o3 (medium reasoning) 24 Sep 2025
UGI: style adherencescore from 0 to 1, higher is better 0.31 331 of 379 o3 (high reasoning) 24 Sep 2025
UGI: style adherencescore from 0 to 1, higher is better 0.28 364 of 379 o3 (low reasoning) 24 Sep 2025
UGI: style adherencescore from 0 to 1, higher is better 0.29 355 of 379 o3 (medium reasoning) 24 Sep 2025
UGI: writing scorescore out of 100, higher is better 66.8 53 of 379 o3 (high reasoning) 24 Sep 2025
UGI: writing scorescore out of 100, higher is better 65.8 58 of 379 o3 (low reasoning) 24 Sep 2025
UGI: writing scorescore out of 100, higher is better 66.8 51 of 379 o3 (medium reasoning) 24 Sep 2025

Not shown: Not other format limits such as character counts or bullet counts.

Test your own prompt.

Benchmarks aren't your task. Run your prompt on the models you could use, with every answer checked and costed.