Model comparison
GPT-4o mini vs o3
o3 is the stronger model overall, scoring 47.5 to 25.5 on the Noometry Index. GPT-4o mini costs 13× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Last verified . 43 shared benchmarks.
Summary
- They share 43 benchmarks with published results for both. GPT-4o mini scores higher in 0 categories and o3 in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where o3 leads 50.2 to 10.4.
- The biggest single-benchmark swing is Aider Polyglot: 3.6% for GPT-4o mini and 81.3% for o3.
- GPT-4o mini is cheaper at $0.15 / $0.60 per million input/output tokens, against $2 / $8 for o3.
- o3 accepts more context: 200K tokens versus 128K.
Side by side
| GPT-4o mini | o3 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 25.5 | 47.5 |
| Released | 2024-07-18 | 2025-04-16 |
| Weights | Proprietary | Proprietary |
| Context window | 128K | 200K |
| Max output | 16K | 100K |
| Input $ / M tokens | $0.15 | $2 |
| Output $ / M tokens | $0.60 | $8 |
| Results tracked | 60 | 63 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o3 leads
GPT-4o mini: 22.0 (#335), o3: 46.8 (#64)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| Aider Polyglot | 3.6% | 81.3% |
| WeirdML | 11.8% | 52.4% |
| LMArena Coding | 1290 | 1408 |
| SWE-bench Verified | — | 62.3% |
| SWE-bench Verified (bash only) | — | 58.4% |
| GSO | — | 8.8% |
| BigCodeBench Instruct | 46.1% | — |
| LiveBench Coding | 43.1% | — |
| BigCodeBench Complete | 57.4% | — |
| CadEval | — | 74% |
| ALE-Bench | — | 933.55 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 72.2% | — |
Agentic & Tool Use o3 leads
GPT-4o mini: 27.5 (#101), o3: 34.5 (#44)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 63% |
| GDPval | — | 30.8% |
| DeepResearch Bench | — | 45.2% |
| OSWorld | — | 23% |
| BALROG | 17.4% | — |
| LMArena Search | — | 1144 |
| METR Time Horizons | — | 65.4% |
Reasoning o3 leads
GPT-4o mini: 8.7 (#347), o3: 32.0 (#78)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| ARC-AGI-2 | 0% | 6.5% |
| SimpleBench | 10.7% | 53.1% |
| Kagi LLM Benchmark | 28.8% | 67.6% |
| Chess Puzzles | 0% | 38% |
| LMArena Hard Prompts | 1267 | 1402 |
| Mystery Game Puzzles | 12% | 29% |
| DTBench | 54.4% | 84.8% |
| LMCA | 10.4% | 39.7% |
| Epoch Capabilities Index | 126.56 | 146.86 |
| ARC-AGI-1 | — | 60.8% |
| CritPt | — | 1.4% |
| EnigmaEval | — | 13.1% |
| LiveBench Reasoning | 32.8% | — |
| LiveBench Data Analysis | 50% | — |
| ForecastBench | — | 62.5 |
| LiveBench | 41.3% | — |
| PIQA | 88.7% | — |
Math o3 leads
GPT-4o mini: 10.4 (#314), o3: 50.2 (#58)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0.7% | 33.3% |
| OTIS Mock AIME 2024-2025 | 6.9% | 84.4% |
| Omni-MATH | 28% | 71.4% |
| LMArena Math | 1267 | 1426 |
| MATH Level 5 | 52.6% | 97.8% |
| LiveBench Math | 36.3% | — |
| FrontierMath (Feb 2025 set) | — | 18.7% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
| GSM8K | 91.3% | — |
Knowledge o3 leads
GPT-4o mini: 17.7 (#284), o3: 54.6 (#52)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| GPQA Diamond | 37.7% | 81.8% |
| SimpleQA Verified | 8.3% | 49.4% |
| MMLU-Pro | 60.3% | 85.9% |
| Confabulations | 37.2% | 14.4% |
| GPQA (HELM) | 36.8% | 75.3% |
| LMArena Expert | 1235 | 1402 |
| Humanity's Last Exam | — | 20.3% |
| BoolQ | 88.7% | — |
| MMLU | 81.8% | — |
Multimodal o3 leads
GPT-4o mini: 25.9 (#122), o3: 41.4 (#36)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| LMArena Vision | 1066 | 1214 |
| GeoBench | 64% | 74% |
| VPCT | 34% | 52% |
| Video-MME | 64.8% | — |
Multilingual o3 leads
GPT-4o mini: 42.0 (#199), o3: 51.7 (#105)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| LMArena Non-English | 1266 | 1401 |
| LMArena Chinese | 1265 | 1437 |
| LMArena French | 1297 | 1430 |
| LMArena German | 1272 | 1420 |
| LMArena Japanese | 1216 | 1403 |
| LMArena Korean | 1195 | 1370 |
| LMArena Russian | 1275 | 1406 |
| LMArena Spanish | 1276 | 1395 |
Instruction Following o3 leads
GPT-4o mini: 61.9 (#239), o3: 72.8 (#127)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| IFEval | 78.2% | 86.9% |
| LMArena Instruction Following | 1258 | 1368 |
| LiveBench Instruction Following | 56.8% | — |
Long Context o3 leads
GPT-4o mini: 39.1 (#186), o3: 53.3 (#6)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| LMArena Longer Query | 1289 | 1372 |
| Fiction.LiveBench | — | 88.9% |
| CL-bench | — | 17.8% |
Writing & Preference o3 leads
GPT-4o mini: 39.5 (#248), o3: 63.5 (#64)
| Benchmark | GPT-4o mini | o3 |
|---|---|---|
| LMArena Text | 1286 | 1410 |
| LMArena Creative Writing | 1268 | 1359 |
| Short-Story Creative Writing | 67.2% | 83.9% |
| EQ-Bench Creative Writing | 873 | 1676 |
| WildBench | 79.1% | 86.1% |
| LMArena Multi-Turn | 1285 | 1405 |
| LiveBench Language | 28.6% | — |
Frequently asked questions
Is GPT-4o mini better than o3?
o3 is the stronger model overall, scoring 47.5 to 25.5 on the Noometry Index. GPT-4o mini costs 13× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Which is cheaper, GPT-4o mini or o3?
GPT-4o mini is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; o3 lists at $2 and $8.
Is GPT-4o mini or o3 better for coding?
o3 scores higher on coding benchmarks: 46.8 versus 22.0 in the Noometry coding category.
Which has the bigger context window?
o3 does, with 200K tokens against 128K.
How many benchmarks do GPT-4o mini and o3 share?
43 benchmarks have published results for both models. GPT-4o mini has 60 scored results on Noometry and o3 has 63.