Model comparison
gpt-oss-20b vs o1
o1 is the stronger model overall, scoring 40.9 to 32.5 on the Noometry Index. gpt-oss-20b costs 729× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Last verified . 23 shared benchmarks.
Summary
- They share 23 benchmarks with published results for both. gpt-oss-20b scores higher in 1 category and o1 in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where o1 leads 55.6 to 35.5.
- The biggest single-benchmark swing is GPQA Diamond: 60.8% for gpt-oss-20b and 76.8% for o1.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $15 / $60 for o1.
- o1 accepts more context: 200K tokens versus 131K.
- gpt-oss-20b has downloadable open weights; the other is API-only.
Side by side
| gpt-oss-20b | o1 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 32.5 | 40.9 |
| Released | 2025-08-05 | 2024-09-12 |
| Weights | Open | Proprietary |
| Context window | 131K | 200K |
| Max output | 16K | 100K |
| Input $ / M tokens | $0.018 | $15 |
| Output $ / M tokens | $0.09 | $60 |
| Results tracked | 34 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o1 leads
gpt-oss-20b: 37.6 (#192), o1: 46.1 (#70)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| WeirdML | 40.9% | 47.6% |
| LMArena Coding | 1306 | 1367 |
| Aider Polyglot | — | 61.7% |
| SciCode | 34.4% | — |
| LiveBench Coding | — | 69.7% |
| CadEval | — | 56% |
| ALE-Bench | 566.05 | — |
| HumanEval+ | — | 89% |
| MBPP+ | — | 80.2% |
Agentic & Tool Use o1 leads
gpt-oss-20b: 9.3 (#154), o1: 24.6 (#117)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| Terminal-Bench | 3.4% | — |
| Cybench | — | 10% |
| METR Time Horizons | — | 51.1% |
Reasoning o1 leads
gpt-oss-20b: 19.3 (#261), o1: 27.9 (#111)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| Chess Puzzles | 4% | 15% |
| LMArena Hard Prompts | 1274 | 1371 |
| DTBench | 68% | 74.7% |
| LMCA | 14.5% | 22.3% |
| Epoch Capabilities Index | 137.82 | 141.91 |
| SimpleBench | — | 41.7% |
| Kagi LLM Benchmark | 53.2% | — |
| ARC-AGI-1 | — | 30.7% |
| CritPt | 1.4% | — |
| EnigmaEval | — | 5.7% |
| LiveBench Reasoning | — | 91.6% |
| LiveBench Data Analysis | — | 65.5% |
| LiveBench | — | 75.7% |
Math gpt-oss-20b leads
gpt-oss-20b: 39.4 (#103), o1: 36.1 (#175)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 65.3% | 73.3% |
| LMArena Math | 1317 | 1388 |
| FrontierMath (Tiers 1-3) | — | 14.7% |
| Omni-MATH | 56.5% | — |
| LiveBench Math | — | 80.3% |
| MATH Level 5 | — | 94.7% |
| FrontierMath (Feb 2025 set) | — | 9.3% |
Knowledge o1 leads
gpt-oss-20b: 34.6 (#195), o1: 41.5 (#110)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| GPQA Diamond | 60.8% | 76.8% |
| LMArena Expert | 1258 | 1361 |
| Humanity's Last Exam | — | 8% |
| SimpleQA Verified | — | 41.1% |
| MMLU-Pro | 74% | — |
| Confabulations | — | 11.7% |
| GPQA (HELM) | 59.4% | — |
Multimodal Not comparable
gpt-oss-20b: —, o1: 34.2 (#93)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| LMArena Vision | — | 1168 |
| GeoBench | — | 80% |
| VPCT | — | 37% |
| SpatialViz-Bench | — | 41.4% |
Multilingual o1 leads
gpt-oss-20b: 42.2 (#197), o1: 48.6 (#142)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| LMArena Non-English | 1268 | 1358 |
| LMArena Chinese | 1314 | 1394 |
| LMArena German | 1255 | 1337 |
| LMArena Japanese | 1244 | 1346 |
| LMArena Korean | 1236 | 1396 |
| LMArena Russian | 1278 | 1356 |
| LMArena Spanish | 1267 | 1345 |
| LMArena French | — | 1344 |
Instruction Following o1 leads
gpt-oss-20b: 61.8 (#240), o1: 74.8 (#86)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| LMArena Instruction Following | 1236 | 1367 |
| LiveBench Instruction Following | — | 81.5% |
| IFEval | 73.2% | — |
Long Context o1 leads
gpt-oss-20b: 37.9 (#209), o1: 50.3 (#9)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| LMArena Longer Query | 1250 | 1378 |
| Fiction.LiveBench | — | 83.3% |
Writing & Preference o1 leads
gpt-oss-20b: 35.5 (#265), o1: 55.6 (#144)
| Benchmark | gpt-oss-20b | o1 |
|---|---|---|
| LMArena Text | 1287 | 1366 |
| LMArena Creative Writing | 1201 | 1348 |
| LMArena Multi-Turn | 1268 | 1369 |
| Short-Story Creative Writing | — | 70.2% |
| EQ-Bench Creative Writing | 666 | — |
| WildBench | 73.7% | — |
| LiveBench Language | — | 65.4% |
Frequently asked questions
Is gpt-oss-20b better than o1?
o1 is the stronger model overall, scoring 40.9 to 32.5 on the Noometry Index. gpt-oss-20b costs 729× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Which is cheaper, gpt-oss-20b or o1?
gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; o1 lists at $15 and $60.
Is gpt-oss-20b or o1 better for coding?
o1 scores higher on coding benchmarks: 46.1 versus 37.6 in the Noometry coding category.
Which has the bigger context window?
o1 does, with 200K tokens against 131K.
How many benchmarks do gpt-oss-20b and o1 share?
23 benchmarks have published results for both models. gpt-oss-20b has 34 scored results on Noometry and o1 has 52.