Model comparison
GPT-4 vs o1-mini
o1-mini is the stronger model overall, scoring 34.0 to 29.1 on the Noometry Index.
Last verified . 23 shared benchmarks.
Summary
- They share 23 benchmarks with published results for both. GPT-4 scores higher in 1 category and o1-mini in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where o1-mini leads 35.4 to 10.8.
- The biggest single-benchmark swing is MATH Level 5: 23% for GPT-4 and 89.2% for o1-mini.
Side by side
| GPT-4 | o1-mini | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 29.1 | 34.0 |
| Released | 2023-03-14 | 2024-09-12 |
| Weights | Proprietary | Proprietary |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o1-mini leads
GPT-4: 31.6 (#283), o1-mini: 35.5 (#224)
| Benchmark | GPT-4 | o1-mini |
|---|---|---|
| WeirdML | 12.4% | 36.3% |
| LMArena Coding | 1254 | 1362 |
| HumanEval+ | 79.3% | 89% |
| Aider Polyglot | — | 32.9% |
| BigCodeBench Instruct | 46% | — |
| LiveBench Coding | — | 48% |
| BigCodeBench Complete | 57.2% | — |
| MBPP+ | — | 78.8% |
Agentic & Tool Use Not comparable
GPT-4: —, o1-mini: 24.6 (#118)
| Benchmark | GPT-4 | o1-mini |
|---|---|---|
| Cybench | — | 10% |
| METR Time Horizons | 36.1% | — |
Reasoning GPT-4 leads
GPT-4: 17.8 (#289), o1-mini: 8.8 (#346)
| Benchmark | GPT-4 | o1-mini |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1333 |
| Epoch Capabilities Index | 125.89 | 135.82 |
| ARC-AGI-2 | — | 0.8% |
| SimpleBench | — | 18.1% |
| ARC-AGI-1 | — | 14% |
| Chess Puzzles | 4% | — |
| LiveBench Reasoning | — | 72.3% |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LiveBench Data Analysis | — | 57.9% |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| LiveBench | — | 57.8% |
| WinoGrande | 87.5% | — |
Math o1-mini leads
GPT-4: 10.8 (#309), o1-mini: 35.4 (#186)
| Benchmark | GPT-4 | o1-mini |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.1% | 46.9% |
| LMArena Math | 1269 | 1358 |
| MATH Level 5 | 23% | 89.2% |
| LiveBench Math | — | 62% |
| FrontierMath (Feb 2025 set) | — | 1.7% |
| GSM8K | 92% | — |
Knowledge o1-mini leads
GPT-4: 18.4 (#282), o1-mini: 34.9 (#192)
| Benchmark | GPT-4 | o1-mini |
|---|---|---|
| GPQA Diamond | 35.7% | 62.4% |
| LMArena Expert | 1211 | 1316 |
| Confabulations | — | 18.6% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multilingual o1-mini leads
GPT-4: 40.6 (#215), o1-mini: 43.6 (#182)
| Benchmark | GPT-4 | o1-mini |
|---|---|---|
| LMArena Non-English | 1246 | 1289 |
| LMArena Chinese | 1242 | 1314 |
| LMArena French | 1283 | 1293 |
| LMArena German | 1251 | 1278 |
| LMArena Japanese | 1209 | 1245 |
| LMArena Korean | 1184 | 1223 |
| LMArena Russian | 1251 | 1283 |
| LMArena Spanish | 1261 | 1303 |
Instruction Following o1-mini leads
GPT-4: 65.3 (#222), o1-mini: 66.7 (#206)
| Benchmark | GPT-4 | o1-mini |
|---|---|---|
| LMArena Instruction Following | 1241 | 1304 |
| LiveBench Instruction Following | — | 65.4% |
Long Context o1-mini leads
GPT-4: 37.7 (#212), o1-mini: 40.1 (#161)
| Benchmark | GPT-4 | o1-mini |
|---|---|---|
| LMArena Longer Query | 1244 | 1320 |
Writing & Preference o1-mini leads
GPT-4: 34.9 (#268), o1-mini: 48.4 (#202)
| Benchmark | GPT-4 | o1-mini |
|---|---|---|
| LMArena Text | 1263 | 1317 |
| LMArena Creative Writing | 1244 | 1244 |
| LMArena Multi-Turn | 1257 | 1314 |
| Short-Story Creative Writing | — | 64.9% |
| EQ-Bench Creative Writing | 752 | — |
| LiveBench Language | — | 40.9% |
Frequently asked questions
Is GPT-4 better than o1-mini?
o1-mini is the stronger model overall, scoring 34.0 to 29.1 on the Noometry Index.
Is GPT-4 or o1-mini better for coding?
o1-mini scores higher on coding benchmarks: 35.5 versus 31.6 in the Noometry coding category.
How many benchmarks do GPT-4 and o1-mini share?
23 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and o1-mini has 39.