Model comparison
GPT-4o vs o1
o1 is the stronger model overall, scoring 40.9 to 28.6 on the Noometry Index. GPT-4o costs 6.0× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Last verified . 51 shared benchmarks.
Summary
- They share 51 benchmarks with published results for both. GPT-4o scores higher in 1 category and o1 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where o1 leads 36.1 to 10.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 6.4% for GPT-4o and 73.3% for o1.
- GPT-4o is cheaper at $2.50 / $10 per million input/output tokens, against $15 / $60 for o1.
- o1 accepts more context: 200K tokens versus 128K.
Side by side
| GPT-4o | o1 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 28.6 | 40.9 |
| Released | 2024-05-13 | 2024-09-12 |
| Weights | Proprietary | Proprietary |
| Context window | 128K | 200K |
| Max output | 16K | 100K |
| Input $ / M tokens | $2.50 | $15 |
| Output $ / M tokens | $10 | $60 |
| Results tracked | 72 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o1 leads
GPT-4o: 24.8 (#328), o1: 46.1 (#70)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| Aider Polyglot | 45.3% | 61.7% |
| WeirdML | 25.1% | 47.6% |
| LiveBench Coding | 51.4% | 69.7% |
| LMArena Coding | 1297 | 1367 |
| CadEval | 26% | 56% |
| HumanEval+ | 87.2% | 89% |
| MBPP+ | 72.2% | 80.2% |
| SWE-bench Verified | 31% | — |
| SWE-bench Verified (bash only) | 21.6% | — |
| GSO | 0% | — |
| BigCodeBench Instruct | 51.1% | — |
| BigCodeBench Complete | 61.1% | — |
Agentic & Tool Use o1 leads
GPT-4o: 21.0 (#141), o1: 24.6 (#117)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| Cybench | 12.5% | 10% |
| METR Time Horizons | 40.8% | 51.1% |
| GDPval | 9.9% | — |
| TheAgentCompany | 8.6% | — |
| BALROG | 32.3% | — |
| LMArena Search | 1006 | — |
Reasoning o1 leads
GPT-4o: 9.4 (#343), o1: 27.9 (#111)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| SimpleBench | 17.8% | 41.7% |
| ARC-AGI-1 | 4.5% | 30.7% |
| Chess Puzzles | 13% | 15% |
| EnigmaEval | 0.8% | 5.7% |
| LiveBench Reasoning | 55.8% | 91.6% |
| LMArena Hard Prompts | 1281 | 1371 |
| DTBench | 64.5% | 74.7% |
| LiveBench Data Analysis | 60.9% | 65.5% |
| LMCA | 16.6% | 22.3% |
| Epoch Capabilities Index | 128.97 | 141.91 |
| LiveBench | 55.3% | 75.7% |
| ARC-AGI-2 | 0% | — |
| CritPt | 0% | — |
| ForecastBench | 57.7 | — |
Math o1 leads
GPT-4o: 10.6 (#312), o1: 36.1 (#175)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0.4% | 14.7% |
| OTIS Mock AIME 2024-2025 | 6.4% | 73.3% |
| LiveBench Math | 49.5% | 80.3% |
| LMArena Math | 1285 | 1388 |
| MATH Level 5 | 53.3% | 94.7% |
| FrontierMath (Feb 2025 set) | 0.3% | 9.3% |
| Omni-MATH | 29.3% | — |
Knowledge o1 leads
GPT-4o: 28.8 (#242), o1: 41.5 (#110)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| GPQA Diamond | 49.2% | 76.8% |
| Humanity's Last Exam | 2.7% | 8% |
| SimpleQA Verified | 26% | 41.1% |
| Confabulations | 15.3% | 11.7% |
| LMArena Expert | 1250 | 1361 |
| MMLU-Pro | 71.3% | — |
| Vectara Hallucination Rate | 9.6% | — |
| GPQA (HELM) | 52% | — |
| MMLU | 88.1% | — |
Multimodal Too close to call
GPT-4o: 34.5 (#91), o1: 34.2 (#93)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| LMArena Vision | 1137 | 1168 |
| GeoBench | 71% | 80% |
| VPCT | 40% | 37% |
| Video-MME | 71.9% | — |
| ScienceQA | 88.5% | — |
| SpatialViz-Bench | — | 41.4% |
Multilingual o1 leads
GPT-4o: 43.2 (#186), o1: 48.6 (#142)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| LMArena Non-English | 1283 | 1358 |
| LMArena Chinese | 1277 | 1394 |
| LMArena French | 1304 | 1344 |
| LMArena German | 1282 | 1337 |
| LMArena Japanese | 1257 | 1346 |
| LMArena Korean | 1234 | 1396 |
| LMArena Russian | 1286 | 1356 |
| LMArena Spanish | 1292 | 1345 |
Instruction Following o1 leads
GPT-4o: 66.6 (#207), o1: 74.8 (#86)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| LiveBench Instruction Following | 68.6% | 81.5% |
| LMArena Instruction Following | 1278 | 1367 |
| IFEval | 81.7% | — |
Long Context o1 leads
GPT-4o: 39.4 (#179), o1: 50.3 (#9)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| Fiction.LiveBench | 66.7% | 83.3% |
| LMArena Longer Query | 1289 | 1378 |
Writing & Preference o1 leads
GPT-4o: 52.6 (#166), o1: 55.6 (#144)
| Benchmark | GPT-4o | o1 |
|---|---|---|
| LMArena Text | 1300 | 1366 |
| LMArena Creative Writing | 1292 | 1348 |
| Short-Story Creative Writing | 81.8% | 70.2% |
| LMArena Multi-Turn | 1302 | 1369 |
| LiveBench Language | 47.6% | 65.4% |
| WildBench | 82.8% | — |
Frequently asked questions
Is GPT-4o better than o1?
o1 is the stronger model overall, scoring 40.9 to 28.6 on the Noometry Index. GPT-4o costs 6.0× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Which is cheaper, GPT-4o or o1?
GPT-4o is cheaper. It lists at $2.50 per million input tokens and $10 per million output tokens; o1 lists at $15 and $60.
Is GPT-4o or o1 better for coding?
o1 scores higher on coding benchmarks: 46.1 versus 24.8 in the Noometry coding category.
Which has the bigger context window?
o1 does, with 200K tokens against 128K.
How many benchmarks do GPT-4o and o1 share?
51 benchmarks have published results for both models. GPT-4o has 72 scored results on Noometry and o1 has 52.