o3 Mini vs o4 Mini
o4 Mini leads o3 Mini on the Artificial Analysis Intelligence Index, 16.7 to 12.5, at about the same output price ($4.40 vs $4.40 per million). Updated October 21, 2026; 68 benchmarks have a result for both.
o3 Mini or o4 Mini: which to pick
Neither model has a coding score yet, so the two cannot be ranked on coding.
Pick o3 Mini for work measured by Aider polyglot or SWE-bench Verified, where it leads.
- Shared results
- 68
- o3 Mini only
- 3
- o4 Mini only
- 30
- Like-for-like categories
- 7 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, six public coding benchmarks
No call
Neither has a coding score yet.
One side not measured
Intelligence
AA Intelligence Index
o4 Mini
o4 Mini leads, 16.7 against 12.5.
Same board, both measured
Reasoning
GPQA Diamond
o3 Mini
o3 Mini leads, 75.5% against 74.5%.
Same board, both measured
Math
AIME
o3 Mini
o3 Mini leads, 86.5% against 83.7%.
Same board, both measured
Knowledge
MMLU Pro
o4 Mini
o4 Mini leads, 80.6% against 78.7%.
Same board, both measured
Web design
Design Arena: websites
No call
Only o4 Mini has a Design Arena: websites result, 985.
Directional only
Long documents
Listed context window
No call
Both take 200K tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
No call
o3 Mini $0.0033, o4 Mini $0.0033 per turn.
List prices
At a glance
- Coding index
- o3 Mini not benchmarked, o4 Mini not benchmarked.
- Widest category
- Agentic work, 24.3 points apart on average over 1 shared benchmark.
- Widest single gap
- Fiction.LiveBench: o3 Mini 50.0% against o4 Mini 77.8%, 27.8 points apart on a 0 to 100 scale.
- Output price
- Both cost $4.40 per million output tokens.
- Context window
- Both take 200K tokens.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | o3 Mini | Bars | o4 Mini | Ties |
|---|---|---|---|---|
| Coding | 1 | 9 | ||
| Intelligence | 0 | 2 | ||
| Agentic work | 0 | 1 | ||
| Reasoning | 5 | 7 | ||
| Math | 3 | 6 | ||
| Knowledge | 0 | 11 | ||
| General | 4 | 19 | ||
| All shared boards | 13 | 55 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index19
| Benchmark | o3 Mini | o4 Mini | Leader |
|---|---|---|---|
| Aider polyglot | 60.4% | 72.0% | o4 Mini |
| SWE-bench Verified | 42.4% | 64.6% | o4 Mini |
| SWE-bench Lite | 32.3% | Not measured | Only o3 Mini measured |
| SWE-bench Multimodal | Not measured | 33.9% | Only o4 Mini measured |
| LiveCodeBench (Vals AI) | 71.5% | 82.2% | o4 Mini |
| LiveCodeBench (Artificial Analysis run) | 73.4% | 85.9% | o4 Mini |
| SciCode | 42.8% | Not measured | Only o3 Mini measured |
| Terminal-Bench Hard (Artificial Analysis run) | 6.8% | 15.2% | o4 Mini |
| GSO-Bench | 1.3% | 3.6% | o4 Mini |
| WeirdML | 43.7% | 52.6% | o4 Mini |
| Cybench | 22.5% | Not measured | Only o3 Mini measured |
| LMArena Text: coding | 1378 | 1368 | o3 Mini |
| LMArena Text: software and IT | 1364 | 1367 | o4 Mini |
| SciCode | 7.7% | 10.8% | o4 Mini |
| Neither model has a result on the other 25 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | o3 Mini | o4 Mini | Leader |
|---|---|---|---|
| AA Intelligence Index | 12.5 | 16.7 | o4 Mini |
| Epoch Capabilities Index | 140.2 | 145.5 | o4 Mini |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | o3 Mini | o4 Mini | Leader |
|---|---|---|---|
| τ²-Bench Telecom | 31.3% | 55.6% | o4 Mini |
| METR time horizons | Not measured | 63.9% | Only o4 Mini measured |
| Berkeley Function Calling Leaderboard (BFCL) | Not measured | 53.2% | Only o4 Mini measured |
| Neither model has a result on the other 38 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index57
| Benchmark | o3 Mini | o4 Mini | Leader |
|---|---|---|---|
| GPQA Diamond | 75.5% | 74.5% | o3 Mini |
| MMMU Pro (Vals AI) | Not measured | 79.7% | Only o4 Mini measured |
| GPQA Diamond (Artificial Analysis run) | 77.3% | 78.4% | o4 Mini |
| Humanity's Last Exam (Artificial Analysis run) | 12.0% | 16.5% | o4 Mini |
| GPQA Diamond (Epoch AI run) | 77.0% | 79.6% | o4 Mini |
| ARC-AGI | 34.5% | 58.7% | o4 Mini |
| ARC-AGI-2 | 3.0% | 6.1% | o4 Mini |
| SimpleBench | 22.8% | 38.7% | o4 Mini |
| Chess puzzles | 15.0% | 22.0% | o4 Mini |
| Mystery game puzzles | 7.0% | 5.0% | o3 Mini |
| VPCT | Not measured | 57.5% | Only o4 Mini measured |
| LMArena Text: hard prompts | 1365 | 1351 | o3 Mini |
| LMArena Text: expert | 1364 | 1343 | o3 Mini |
| LMArena Text: hard prompts in English | 1375 | 1366 | o3 Mini |
| LMArena Vision: diagrams | Not measured | 1197 | Only o4 Mini measured |
| LMArena Vision: homework | Not measured | 1234 | Only o4 Mini measured |
| Kagi LLM Benchmark | Not measured | 67.6% | Only o4 Mini measured |
| Neither model has a result on the other 5 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index36
| Benchmark | o3 Mini | o4 Mini | Leader |
|---|---|---|---|
| AIME | 86.5% | 83.7% | o3 Mini |
| MATH 500 (Vals AI) | 91.8% | 94.2% | o4 Mini |
| MGSM (Vals AI) | 91.3% | 93.4% | o4 Mini |
| AA Math Index | Not measured | 90.7 | Only o4 Mini measured |
| AIME 2025 (Artificial Analysis run) | Not measured | 90.7% | Only o4 Mini measured |
| OTIS Mock AIME | 76.9% | 81.7% | o4 Mini |
| FrontierMath, tiers 1 to 3 | 18.6% | 36.1% | o4 Mini |
| FrontierMath, tier 4 | 0.0% | 4.9% | o4 Mini |
| MATH level 5 | 96.5% | 97.8% | o4 Mini |
| LMArena Text: math | 1396 | 1389 | o3 Mini |
| LMArena Text: mathematical work | 1399 | 1379 | o3 Mini |
| Neither model has a result on the other 2 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index011
| Benchmark | o3 Mini | o4 Mini | Leader |
|---|---|---|---|
| MMLU Pro | 78.7% | 80.6% | o4 Mini |
| LegalBench (Vals AI) | 71.5% | 79.2% | o4 Mini |
| CorpFin (Vals AI) | 45.3% | 59.0% | o4 Mini |
| TaxEval (Vals AI) | 69.4% | 74.8% | o4 Mini |
| MortgageTax (Vals AI) | Not measured | 64.8% | Only o4 Mini measured |
| MedQA (Vals AI) | 94.8% | 96.0% | o4 Mini |
| MedCode (Vals AI) | Not measured | 33.8% | Only o4 Mini measured |
| MMLU Pro (Artificial Analysis run) | 80.2% | 83.2% | o4 Mini |
| SimpleQA Verified | 15.3% | 19.6% | o4 Mini |
| GeoBench | Not measured | 64.0% | Only o4 Mini measured |
| LMArena Text: science | 1339 | 1354 | o4 Mini |
| LMArena Text: medicine and healthcare | 1330 | 1361 | o4 Mini |
| LMArena Text: legal and government | 1328 | 1350 | o4 Mini |
| LMArena Text: business and finance | 1313 | 1325 | o4 Mini |
| FACTS Benchmark Suite | Not measured | 28.1% | Only o4 Mini measured |
| Neither model has a result on the other 3 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index6 boards, none shared
| Benchmark | o3 Mini | o4 Mini | Leader |
|---|---|---|---|
| Design Arena: websites | Not measured | 985 | Only o4 Mini measured |
| Design Arena: UI components | Not measured | 983 | Only o4 Mini measured |
| Design Arena: game dev | Not measured | 1003 | Only o4 Mini measured |
| Design Arena: data viz | Not measured | 990 | Only o4 Mini measured |
| Design Arena: 3D | Not measured | 854 | Only o4 Mini measured |
| Design Arena: all code categories | Not measured | 976 | Only o4 Mini measured |
| Neither model has a result on the other 3 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index419
| Benchmark | o3 Mini | o4 Mini | Leader |
|---|---|---|---|
| SAGE (Vals AI) | Not measured | 41.1% | Only o4 Mini measured |
| MedScribe (Vals AI) | Not measured | 69.1% | Only o4 Mini measured |
| IFBench | 67.1% | 68.7% | o4 Mini |
| Long context reasoning (AA-LCR) | 43.0% | 61.0% | o4 Mini |
| DTBench | 68.8% | 77.6% | o4 Mini |
| LMCA | 19.0% | 26.5% | o4 Mini |
| Fiction.LiveBench | 50.0% | 77.8% | o4 Mini |
| Creative writing (Lech Mazur) | 61.7% | 75.0% | o4 Mini |
| LMArena Text | 1337 | 1353 | o4 Mini |
| LMArena Text: instruction following | 1337 | 1320 | o3 Mini |
| LMArena Text: creative writing | 1286 | 1294 | o4 Mini |
| LMArena Vision | Not measured | 1194 | Only o4 Mini measured |
| LMArena Text: English | 1351 | 1367 | o4 Mini |
| LMArena Text: non-English | 1319 | 1337 | o4 Mini |
| LMArena Text: Chinese | 1379 | 1354 | o3 Mini |
| LMArena Text: German | 1303 | 1336 | o4 Mini |
| LMArena Text: Japanese | 1286 | 1308 | o4 Mini |
| LMArena Text: Korean | 1313 | 1312 | o3 Mini |
| LMArena Text: French | 1334 | 1364 | o4 Mini |
| LMArena Text: Spanish | 1321 | 1346 | o4 Mini |
| LMArena Text: Russian | 1304 | 1334 | o4 Mini |
| LMArena Text: Polish | 1294 | 1358 | o4 Mini |
| LMArena Text: multi-turn | 1320 | 1350 | o4 Mini |
| LMArena Text: longer queries | 1343 | 1315 | o3 Mini |
| LMArena Text: writing and language | 1313 | 1316 | o4 Mini |
| LMArena Text: entertainment and media | 1283 | 1308 | o4 Mini |
| LMArena Vision: English | Not measured | 1200 | Only o4 Mini measured |
| LMArena Vision: Chinese | Not measured | 1218 | Only o4 Mini measured |
| LMArena Vision: reading text | Not measured | 1197 | Only o4 Mini measured |
| LMArena Vision: humor | Not measured | 1199 | Only o4 Mini measured |
| LMArena Vision: captioning | Not measured | 1204 | Only o4 Mini measured |
| LMArena Vision: recognizing things | Not measured | 1218 | Only o4 Mini measured |
| LMArena Vision: creative writing | Not measured | 1166 | Only o4 Mini measured |
| Neither model has a result on the other 13 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | o3 Mini | o4 Mini | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Same cost
Repository review
50K input + 3K output tokens
Same cost
Agent loop
200K input + 10K output tokens
Same cost
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | OpenAI o3 Mini | OpenAI o4 Mini |
|---|---|---|
| Maker | OpenAI | OpenAI |
| Weights | No public weights listed | No public weights listed |
| Released | January 31, 2025 | April 16, 2025 |
| Context window | 200K | 200K |
| Input $/Mtok | $1.10 | $1.10 |
| Output $/Mtok | $4.40 | $4.40 |
| Coding index | Not benchmarked | Not benchmarked |
| Independent measurements behind it | 0 | 0 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is o3 Mini better than o4 Mini for coding?
Neither model has a coding score yet, so the two cannot be ranked on coding. Pick o3 Mini for work measured by Aider polyglot or SWE-bench Verified, where it leads.
Which is cheaper, o3 Mini or o4 Mini?
o3 Mini costs $1.10 per million input tokens and $4.40 per million output tokens; o4 Mini costs $1.10 and $4.40. Cost per solved task is known for neither of them.
Which has the bigger context window, o3 Mini or o4 Mini?
Neither: both take 200K tokens.
How many benchmarks is the o3 Mini vs o4 Mini comparison based on?
68 benchmarks have a result for both models as of October 21, 2026. o3 Mini's coding index rests on 0 independent benchmarks and o4 Mini's on 0. Where only one model has a result, the page says so instead of naming a winner.