Model comparison
GPT-4 vs o1
o1 is the stronger model overall, scoring 40.9 to 29.1 on the Noometry Index.
Last verified . 27 shared benchmarks.
Summary
- They share 27 benchmarks with published results for both. GPT-4 scores higher in 0 categories and o1 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where o1 leads 36.1 to 10.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.1% for GPT-4 and 73.3% for o1.
- o1 is cheaper at $15 / $60 per million input/output tokens, against $30 / $60 for GPT-4.
- o1 accepts more context: 200K tokens versus 8K.
Side by side
| GPT-4 | o1 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 29.1 | 40.9 |
| Released | 2023-03-14 | 2024-09-12 |
| Weights | Proprietary | Proprietary |
| Context window | 8K | 200K |
| Max output | 8K | 100K |
| Input $ / M tokens | $30 | $15 |
| Output $ / M tokens | $60 | $60 |
| Results tracked | 38 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o1 leads
GPT-4: 31.6 (#283), o1: 46.1 (#70)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| WeirdML | 12.4% | 47.6% |
| LMArena Coding | 1254 | 1367 |
| HumanEval+ | 79.3% | 89% |
| Aider Polyglot | — | 61.7% |
| BigCodeBench Instruct | 46% | — |
| LiveBench Coding | — | 69.7% |
| BigCodeBench Complete | 57.2% | — |
| CadEval | — | 56% |
| MBPP+ | — | 80.2% |
Agentic & Tool Use Not comparable
GPT-4: —, o1: 24.6 (#117)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| METR Time Horizons | 36.1% | 51.1% |
| Cybench | — | 10% |
Reasoning o1 leads
GPT-4: 17.8 (#289), o1: 27.9 (#111)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| Chess Puzzles | 4% | 15% |
| LMArena Hard Prompts | 1241 | 1371 |
| DTBench | 62.7% | 74.7% |
| LMCA | 17.1% | 22.3% |
| Epoch Capabilities Index | 125.89 | 141.91 |
| SimpleBench | — | 41.7% |
| ARC-AGI-1 | — | 30.7% |
| EnigmaEval | — | 5.7% |
| LiveBench Reasoning | — | 91.6% |
| Mystery Game Puzzles | 12% | — |
| LiveBench Data Analysis | — | 65.5% |
| BIG-Bench Hard | 75.1% | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| LiveBench | — | 75.7% |
| WinoGrande | 87.5% | — |
Math o1 leads
GPT-4: 10.8 (#309), o1: 36.1 (#175)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.1% | 73.3% |
| LMArena Math | 1269 | 1388 |
| MATH Level 5 | 23% | 94.7% |
| FrontierMath (Tiers 1-3) | — | 14.7% |
| LiveBench Math | — | 80.3% |
| FrontierMath (Feb 2025 set) | — | 9.3% |
| GSM8K | 92% | — |
Knowledge o1 leads
GPT-4: 18.4 (#282), o1: 41.5 (#110)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| GPQA Diamond | 35.7% | 76.8% |
| LMArena Expert | 1211 | 1361 |
| Humanity's Last Exam | — | 8% |
| SimpleQA Verified | — | 41.1% |
| Confabulations | — | 11.7% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multimodal Not comparable
GPT-4: —, o1: 34.2 (#93)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| LMArena Vision | — | 1168 |
| GeoBench | — | 80% |
| VPCT | — | 37% |
| SpatialViz-Bench | — | 41.4% |
Multilingual o1 leads
GPT-4: 40.6 (#215), o1: 48.6 (#142)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| LMArena Non-English | 1246 | 1358 |
| LMArena Chinese | 1242 | 1394 |
| LMArena French | 1283 | 1344 |
| LMArena German | 1251 | 1337 |
| LMArena Japanese | 1209 | 1346 |
| LMArena Korean | 1184 | 1396 |
| LMArena Russian | 1251 | 1356 |
| LMArena Spanish | 1261 | 1345 |
Instruction Following o1 leads
GPT-4: 65.3 (#222), o1: 74.8 (#86)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| LMArena Instruction Following | 1241 | 1367 |
| LiveBench Instruction Following | — | 81.5% |
Long Context o1 leads
GPT-4: 37.7 (#212), o1: 50.3 (#9)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| LMArena Longer Query | 1244 | 1378 |
| Fiction.LiveBench | — | 83.3% |
Writing & Preference o1 leads
GPT-4: 34.9 (#268), o1: 55.6 (#144)
| Benchmark | GPT-4 | o1 |
|---|---|---|
| LMArena Text | 1263 | 1366 |
| LMArena Creative Writing | 1244 | 1348 |
| LMArena Multi-Turn | 1257 | 1369 |
| Short-Story Creative Writing | — | 70.2% |
| EQ-Bench Creative Writing | 752 | — |
| LiveBench Language | — | 65.4% |
Frequently asked questions
Is GPT-4 better than o1?
o1 is the stronger model overall, scoring 40.9 to 29.1 on the Noometry Index.
Which is cheaper, GPT-4 or o1?
o1 is cheaper. It lists at $15 per million input tokens and $60 per million output tokens; GPT-4 lists at $30 and $60.
Is GPT-4 or o1 better for coding?
o1 scores higher on coding benchmarks: 46.1 versus 31.6 in the Noometry coding category.
Which has the bigger context window?
o1 does, with 200K tokens against 8K.
How many benchmarks do GPT-4 and o1 share?
27 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and o1 has 52.