Model comparison
Claude Opus 4 vs gpt-oss-120b
Claude Opus 4 is the stronger model overall, scoring 43.1 to 36.3 on the Noometry Index. gpt-oss-120b costs 427× less per token, which makes it the better buy when Claude Opus 4's lead doesn't matter for your workload.
Last verified . 40 shared benchmarks.
Summary
- They share 40 benchmarks with published results for both. Claude Opus 4 scores higher in 8 categories and gpt-oss-120b in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where Claude Opus 4 leads 34.8 to 12.2.
- The biggest single-benchmark swing is SWE-bench Verified (bash only): 67.6% for Claude Opus 4 and 26% for gpt-oss-120b.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $15 / $75 for Claude Opus 4.
- Claude Opus 4 accepts more context: 200K tokens versus 131K.
- gpt-oss-120b has downloadable open weights; the other is API-only.
Side by side
| Claude Opus 4 | gpt-oss-120b | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 43.1 | 36.3 |
| Released | 2025-05-22 | 2025-08-05 |
| Weights | Proprietary | Open |
| Context window | 200K | 131K |
| Max output | 32K | 41K |
| Input $ / M tokens | $15 | $0.037 |
| Output $ / M tokens | $75 | $0.17 |
| Results tracked | 56 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 4 leads
Claude Opus 4: 47.2 (#62), gpt-oss-120b: 33.5 (#256)
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| SWE-bench Verified (bash only) | 67.6% | 26% |
| Aider Polyglot | 72% | 41.8% |
| WeirdML | 43.7% | 48.2% |
| LMArena Coding | 1442 | 1380 |
| AlgoTune | 1.33 | 1.41 |
| SWE-bench Verified | 70.7% | — |
| SciCode | — | 36% |
| GSO | 6.9% | — |
| ALE-Bench | — | 575.62 |
Agentic & Tool Use Claude Opus 4 leads
Claude Opus 4: 34.8 (#42), gpt-oss-120b: 12.2 (#153)
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| METR Time Horizons | 63.9% | 56.6% |
| Terminal-Bench | — | 18.7% |
| APEX-Agents | — | 4.4% |
| Cybench | 38% | — |
| DeepResearch Bench | 46.8% | — |
| LMArena Search | 1127 | — |
| Vending-Bench 2 | — | -21.53 |
Reasoning Claude Opus 4 leads
Claude Opus 4: 27.3 (#121), gpt-oss-120b: 20.0 (#245)
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| SimpleBench | 58.8% | 22.1% |
| Kagi LLM Benchmark | 74.3% | 58.6% |
| CritPt | 0.3% | 1.1% |
| LMArena Hard Prompts | 1399 | 1364 |
| DTBench | 81.6% | 76.3% |
| LMCA | 37.4% | 22.1% |
| Epoch Capabilities Index | 142.67 | 139.93 |
| ARC-AGI-2 | 8.6% | — |
| ARC-AGI-1 | 35.7% | — |
| Chess Puzzles | — | 20% |
| EnigmaEval | 5.6% | — |
| Mystery Game Puzzles | — | 2% |
| Surface Evolver Bench | — | 25% |
| ForecastBench | 61.1 | — |
Math gpt-oss-120b leads
Claude Opus 4: 42.0 (#86), gpt-oss-120b: 52.5 (#50)
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 64.4% | 88.9% |
| Omni-MATH | 61.6% | 68.8% |
| LMArena Math | 1390 | 1389 |
| MATH Level 5 | 85% | — |
| FrontierMath (Feb 2025 set) | 4.5% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |
Knowledge Claude Opus 4 leads
Claude Opus 4: 44.0 (#88), gpt-oss-120b: 42.4 (#96)
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| GPQA Diamond | 76.3% | 75.8% |
| MMLU-Pro | 87.5% | 79.5% |
| Confabulations | 15.9% | 15.7% |
| Vectara Hallucination Rate | 12% | 14.2% |
| GPQA (HELM) | 70.8% | 68.4% |
| LMArena Expert | 1386 | 1356 |
| Humanity's Last Exam | 10.7% | — |
Multimodal Not comparable
Claude Opus 4: 31.5 (#106), gpt-oss-120b: —
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| LMArena Vision | 1192 | — |
| GeoBench | 49% | — |
| VPCT | 38% | — |
Multilingual Too close to call
Claude Opus 4: 48.8 (#138), gpt-oss-120b: 48.0 (#147)
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| LMArena Non-English | 1362 | 1351 |
| LMArena Chinese | 1386 | 1385 |
| LMArena French | 1372 | 1369 |
| LMArena German | 1391 | 1353 |
| LMArena Japanese | 1331 | 1331 |
| LMArena Korean | 1321 | 1282 |
| LMArena Russian | 1392 | 1343 |
| LMArena Spanish | 1389 | 1389 |
Instruction Following Claude Opus 4 leads
Claude Opus 4: 77.1 (#28), gpt-oss-120b: 69.3 (#173)
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| IFEval | 91.8% | 83.6% |
| LMArena Instruction Following | 1406 | 1318 |
Long Context Claude Opus 4 leads
Claude Opus 4: 39.6 (#172), gpt-oss-120b: 31.4 (#278)
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| Fiction.LiveBench | 61.1% | 44.4% |
| LMArena Longer Query | 1422 | 1319 |
Writing & Preference Claude Opus 4 leads
Claude Opus 4: 61.2 (#89), gpt-oss-120b: 46.5 (#217)
| Benchmark | Claude Opus 4 | gpt-oss-120b |
|---|---|---|
| LMArena Text | 1377 | 1365 |
| LMArena Creative Writing | 1387 | 1275 |
| Short-Story Creative Writing | 83.6% | 77.1% |
| EQ-Bench Creative Writing | 1580 | 961 |
| WildBench | 85.2% | 84.5% |
| LMArena Multi-Turn | 1396 | 1340 |
Frequently asked questions
Is Claude Opus 4 better than gpt-oss-120b?
Claude Opus 4 is the stronger model overall, scoring 43.1 to 36.3 on the Noometry Index. gpt-oss-120b costs 427× less per token, which makes it the better buy when Claude Opus 4's lead doesn't matter for your workload.
Which is cheaper, Claude Opus 4 or gpt-oss-120b?
gpt-oss-120b is cheaper. It lists at $0.037 per million input tokens and $0.17 per million output tokens; Claude Opus 4 lists at $15 and $75.
Is Claude Opus 4 or gpt-oss-120b better for coding?
Claude Opus 4 scores higher on coding benchmarks: 47.2 versus 33.5 in the Noometry coding category.
Which has the bigger context window?
Claude Opus 4 does, with 200K tokens against 131K.
How many benchmarks do Claude Opus 4 and gpt-oss-120b share?
40 benchmarks have published results for both models. Claude Opus 4 has 56 scored results on Noometry and gpt-oss-120b has 48.