Model comparison
gpt-oss-120b vs o1
o1 is the stronger model overall, scoring 40.9 to 36.3 on the Noometry Index. gpt-oss-120b costs 374× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Last verified . 30 shared benchmarks.
Summary
- They share 30 benchmarks with published results for both. gpt-oss-120b scores higher in 2 categories and o1 in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in long context, where o1 leads 50.3 to 31.4.
- The biggest single-benchmark swing is Fiction.LiveBench: 44.4% for gpt-oss-120b and 83.3% for o1.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $15 / $60 for o1.
- o1 accepts more context: 200K tokens versus 131K.
- gpt-oss-120b has downloadable open weights; the other is API-only.
Side by side
| gpt-oss-120b | o1 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 36.3 | 40.9 |
| Released | 2025-08-05 | 2024-09-12 |
| Weights | Open | Proprietary |
| Context window | 131K | 200K |
| Max output | 41K | 100K |
| Input $ / M tokens | $0.037 | $15 |
| Output $ / M tokens | $0.17 | $60 |
| Results tracked | 48 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o1 leads
gpt-oss-120b: 33.5 (#256), o1: 46.1 (#70)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| Aider Polyglot | 41.8% | 61.7% |
| WeirdML | 48.2% | 47.6% |
| LMArena Coding | 1380 | 1367 |
| SWE-bench Verified (bash only) | 26% | — |
| SciCode | 36% | — |
| LiveBench Coding | — | 69.7% |
| CadEval | — | 56% |
| ALE-Bench | 575.62 | — |
| AlgoTune | 1.41 | — |
| HumanEval+ | — | 89% |
| MBPP+ | — | 80.2% |
Agentic & Tool Use o1 leads
gpt-oss-120b: 12.2 (#153), o1: 24.6 (#117)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| METR Time Horizons | 56.6% | 51.1% |
| Terminal-Bench | 18.7% | — |
| APEX-Agents | 4.4% | — |
| Cybench | — | 10% |
| Vending-Bench 2 | -21.53 | — |
Reasoning o1 leads
gpt-oss-120b: 20.0 (#245), o1: 27.9 (#111)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| SimpleBench | 22.1% | 41.7% |
| Chess Puzzles | 20% | 15% |
| LMArena Hard Prompts | 1364 | 1371 |
| DTBench | 76.3% | 74.7% |
| LMCA | 22.1% | 22.3% |
| Epoch Capabilities Index | 139.93 | 141.91 |
| Kagi LLM Benchmark | 58.6% | — |
| ARC-AGI-1 | — | 30.7% |
| CritPt | 1.1% | — |
| EnigmaEval | — | 5.7% |
| LiveBench Reasoning | — | 91.6% |
| Mystery Game Puzzles | 2% | — |
| LiveBench Data Analysis | — | 65.5% |
| Surface Evolver Bench | 25% | — |
| LiveBench | — | 75.7% |
Math gpt-oss-120b leads
gpt-oss-120b: 52.5 (#50), o1: 36.1 (#175)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 73.3% |
| LMArena Math | 1389 | 1388 |
| FrontierMath (Tiers 1-3) | — | 14.7% |
| Omni-MATH | 68.8% | — |
| LiveBench Math | — | 80.3% |
| MATH Level 5 | — | 94.7% |
| FrontierMath (Feb 2025 set) | — | 9.3% |
Knowledge Too close to call
gpt-oss-120b: 42.4 (#96), o1: 41.5 (#110)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| GPQA Diamond | 75.8% | 76.8% |
| Confabulations | 15.7% | 11.7% |
| LMArena Expert | 1356 | 1361 |
| Humanity's Last Exam | — | 8% |
| SimpleQA Verified | — | 41.1% |
| MMLU-Pro | 79.5% | — |
| Vectara Hallucination Rate | 14.2% | — |
| GPQA (HELM) | 68.4% | — |
Multimodal Not comparable
gpt-oss-120b: —, o1: 34.2 (#93)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| LMArena Vision | — | 1168 |
| GeoBench | — | 80% |
| VPCT | — | 37% |
| SpatialViz-Bench | — | 41.4% |
Multilingual Too close to call
gpt-oss-120b: 48.0 (#147), o1: 48.6 (#142)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| LMArena Non-English | 1351 | 1358 |
| LMArena Chinese | 1385 | 1394 |
| LMArena French | 1369 | 1344 |
| LMArena German | 1353 | 1337 |
| LMArena Japanese | 1331 | 1346 |
| LMArena Korean | 1282 | 1396 |
| LMArena Russian | 1343 | 1356 |
| LMArena Spanish | 1389 | 1345 |
Instruction Following o1 leads
gpt-oss-120b: 69.3 (#173), o1: 74.8 (#86)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| LMArena Instruction Following | 1318 | 1367 |
| LiveBench Instruction Following | — | 81.5% |
| IFEval | 83.6% | — |
Long Context o1 leads
gpt-oss-120b: 31.4 (#278), o1: 50.3 (#9)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| Fiction.LiveBench | 44.4% | 83.3% |
| LMArena Longer Query | 1319 | 1378 |
Writing & Preference o1 leads
gpt-oss-120b: 46.5 (#217), o1: 55.6 (#144)
| Benchmark | gpt-oss-120b | o1 |
|---|---|---|
| LMArena Text | 1365 | 1366 |
| LMArena Creative Writing | 1275 | 1348 |
| Short-Story Creative Writing | 77.1% | 70.2% |
| LMArena Multi-Turn | 1340 | 1369 |
| EQ-Bench Creative Writing | 961 | — |
| WildBench | 84.5% | — |
| LiveBench Language | — | 65.4% |
Frequently asked questions
Is gpt-oss-120b better than o1?
o1 is the stronger model overall, scoring 40.9 to 36.3 on the Noometry Index. gpt-oss-120b costs 374× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Which is cheaper, gpt-oss-120b or o1?
gpt-oss-120b is cheaper. It lists at $0.037 per million input tokens and $0.17 per million output tokens; o1 lists at $15 and $60.
Is gpt-oss-120b or o1 better for coding?
o1 scores higher on coding benchmarks: 46.1 versus 33.5 in the Noometry coding category.
Which has the bigger context window?
o1 does, with 200K tokens against 131K.
How many benchmarks do gpt-oss-120b and o1 share?
30 benchmarks have published results for both models. gpt-oss-120b has 48 scored results on Noometry and o1 has 52.