Model comparison
o3 vs o4-mini
o3 is the stronger model overall, scoring 47.5 to 41.6 on the Noometry Index. o4-mini costs 1.8× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Last verified . 57 shared benchmarks.
Summary
- They share 57 benchmarks with published results for both. o3 scores higher in 9 categories and o4-mini in 1 category; 10 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where o3 leads 54.6 to 43.6.
- The biggest single-benchmark swing is SimpleQA Verified: 49.4% for o3 and 19.6% for o4-mini.
- o4-mini is cheaper at $1.10 / $4.40 per million input/output tokens, against $2 / $8 for o3.
Side by side
| o3 | o4-mini | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 47.5 | 41.6 |
| Released | 2025-04-16 | 2025-04-16 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 200K |
| Max output | 100K | 100K |
| Input $ / M tokens | $2 | $1.10 |
| Output $ / M tokens | $8 | $4.40 |
| Results tracked | 63 | 60 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o3 leads
o3: 46.8 (#64), o4-mini: 40.9 (#127)
| Benchmark | o3 | o4-mini |
|---|---|---|
| SWE-bench Verified (bash only) | 58.4% | 45% |
| Aider Polyglot | 81.3% | 72% |
| GSO | 8.8% | 3.6% |
| WeirdML | 52.4% | 52.6% |
| LMArena Coding | 1408 | 1368 |
| CadEval | 74% | 62% |
| ALE-Bench | 933.55 | 826.17 |
| SWE-bench Verified | 62.3% | — |
| AlgoTune | — | 1.72 |
Agentic & Tool Use o3 leads
o3: 34.5 (#44), o4-mini: 32.6 (#61)
| Benchmark | o3 | o4-mini |
|---|---|---|
| Berkeley Function Calling Leaderboard | 63% | 53.2% |
| GDPval | 30.8% | 25.3% |
| METR Time Horizons | 65.4% | 63.9% |
| DeepResearch Bench | 45.2% | — |
| OSWorld | 23% | — |
| LMArena Search | 1144 | — |
Reasoning o3 leads
o3: 32.0 (#78), o4-mini: 24.6 (#162)
| Benchmark | o3 | o4-mini |
|---|---|---|
| ARC-AGI-2 | 6.5% | 6.1% |
| SimpleBench | 53.1% | 38.7% |
| Kagi LLM Benchmark | 67.6% | 67.6% |
| ARC-AGI-1 | 60.8% | 58.7% |
| CritPt | 1.4% | 0.6% |
| Chess Puzzles | 38% | 26% |
| EnigmaEval | 13.1% | 9.2% |
| LMArena Hard Prompts | 1402 | 1351 |
| Mystery Game Puzzles | 29% | 5% |
| DTBench | 84.8% | 77.6% |
| LMCA | 39.7% | 26.5% |
| Epoch Capabilities Index | 146.86 | 145.64 |
| ForecastBench | 62.5 | 61.8 |
Math o3 leads
o3: 50.2 (#58), o4-mini: 40.8 (#89)
| Benchmark | o3 | o4-mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | 33.3% | 36.1% |
| OTIS Mock AIME 2024-2025 | 84.4% | 81.7% |
| Omni-MATH | 71.4% | 72% |
| LMArena Math | 1426 | 1389 |
| MATH Level 5 | 97.8% | 97.8% |
| FrontierMath (Feb 2025 set) | 18.7% | 24.8% |
| FrontierMath Tier 4 (v1) | 2.1% | 6.3% |
| FrontierMath Tier 4 | — | 4.9% |
Knowledge o3 leads
o3: 54.6 (#52), o4-mini: 43.6 (#91)
| Benchmark | o3 | o4-mini |
|---|---|---|
| GPQA Diamond | 81.8% | 79.6% |
| Humanity's Last Exam | 20.3% | 18.1% |
| SimpleQA Verified | 49.4% | 19.6% |
| MMLU-Pro | 85.9% | 82% |
| Confabulations | 14.4% | 15.8% |
| GPQA (HELM) | 75.3% | 73.5% |
| LMArena Expert | 1402 | 1343 |
| Vectara Hallucination Rate | — | 18.6% |
Multimodal o3 leads
o3: 41.4 (#36), o4-mini: 40.2 (#49)
| Benchmark | o3 | o4-mini |
|---|---|---|
| LMArena Vision | 1214 | 1194 |
| GeoBench | 74% | 64% |
| VPCT | 52% | 57.5% |
Multilingual o3 leads
o3: 51.7 (#105), o4-mini: 47.0 (#154)
| Benchmark | o3 | o4-mini |
|---|---|---|
| LMArena Non-English | 1401 | 1337 |
| LMArena Chinese | 1437 | 1354 |
| LMArena French | 1430 | 1364 |
| LMArena German | 1420 | 1336 |
| LMArena Japanese | 1403 | 1308 |
| LMArena Korean | 1370 | 1312 |
| LMArena Russian | 1406 | 1334 |
| LMArena Spanish | 1395 | 1347 |
Instruction Following o4-mini leads
o3: 72.8 (#127), o4-mini: 75.2 (#68)
| Benchmark | o3 | o4-mini |
|---|---|---|
| IFEval | 86.9% | 92.8% |
| LMArena Instruction Following | 1368 | 1321 |
Long Context o3 leads
o3: 53.3 (#6), o4-mini: 45.5 (#33)
| Benchmark | o3 | o4-mini |
|---|---|---|
| Fiction.LiveBench | 88.9% | 77.8% |
| LMArena Longer Query | 1372 | 1315 |
| CL-bench | 17.8% | — |
Writing & Preference o3 leads
o3: 63.5 (#64), o4-mini: 54.0 (#152)
| Benchmark | o3 | o4-mini |
|---|---|---|
| LMArena Text | 1410 | 1353 |
| LMArena Creative Writing | 1359 | 1294 |
| Short-Story Creative Writing | 83.9% | 75% |
| WildBench | 86.1% | 85.4% |
| LMArena Multi-Turn | 1405 | 1350 |
| EQ-Bench Creative Writing | 1676 | — |
Frequently asked questions
Is o3 better than o4-mini?
o3 is the stronger model overall, scoring 47.5 to 41.6 on the Noometry Index. o4-mini costs 1.8× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Which is cheaper, o3 or o4-mini?
o4-mini is cheaper. It lists at $1.10 per million input tokens and $4.40 per million output tokens; o3 lists at $2 and $8.
Is o3 or o4-mini better for coding?
o3 scores higher on coding benchmarks: 46.8 versus 40.9 in the Noometry coding category.
Which has the bigger context window?
Both accept 200K tokens.
How many benchmarks do o3 and o4-mini share?
57 benchmarks have published results for both models. o3 has 63 scored results on Noometry and o4-mini has 60.