Model comparison
gpt-oss-120b vs Qwen2.5 72B Instruct
gpt-oss-120b is the stronger model overall, scoring 36.0 to 31.5 on the Noometry Index.
Last verified . 30 shared benchmarks.
Alibaba (Qwen)Summary
- They share 30 benchmarks with published results for both. gpt-oss-120b scores higher in 5 categories and Qwen2.5 72B Instruct in 4 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where gpt-oss-120b leads 50.6 to 18.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 88.9% for gpt-oss-120b and 8.1% for Qwen2.5 72B Instruct.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
Side by side
| gpt-oss-120b | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 36.0 | 31.5 |
| Released | 2025-08-05 | 2024-09 |
| Weights | Open | Open |
| Context window | 131K | 131K |
| Max output | 41K | 8K |
| Input $ / M tokens | $0.037 | $1.40 |
| Output $ / M tokens | $0.17 | $5.60 |
| Results tracked | 48 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
gpt-oss-120b: 33.2 (#251), Qwen2.5 72B Instruct: 32.8 (#256)
| Benchmark | gpt-oss-120b | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | 48.2% | 16% |
| LMArena Coding | 1380 | 1293 |
| SWE-bench Verified (bash only) | 26% | — |
| Aider Polyglot | 41.8% | — |
| SciCode | 36% | — |
| BigCodeBench Instruct | — | 45.8% |
| BigCodeBench Complete | — | 55.9% |
| ALE-Bench | 575.62 | — |
| AlgoTune | 1.41 | — |
Agentic & Tool Use Qwen2.5 72B Instruct leads
gpt-oss-120b: 12.1 (#145), Qwen2.5 72B Instruct: 22.0 (#124)
| Benchmark | gpt-oss-120b | Qwen2.5 72B Instruct |
|---|---|---|
| METR Time Horizons | 56.6% | 35.8% |
| Terminal-Bench | 18.7% | — |
| APEX-Agents | 4.4% | — |
| TheAgentCompany | — | 5.7% |
| BALROG | — | 16.2% |
| Vending-Bench 2 | -21.53 | — |
Reasoning Qwen2.5 72B Instruct leads
gpt-oss-120b: 19.5 (#236), Qwen2.5 72B Instruct: 21.4 (#197)
| Benchmark | gpt-oss-120b | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1364 | 1271 |
| DTBench | 76.3% | 62.9% |
| LMCA | 22.1% | 13.4% |
| Epoch Capabilities Index | 139.94 | 129.01 |
| SimpleBench | 22.1% | — |
| Kagi LLM Benchmark | 58.6% | — |
| CritPt | 1.1% | — |
| Chess Puzzles | 20% | — |
| Mystery Game Puzzles | 2% | — |
| Surface Evolver Bench | 25% | — |
| BIG-Bench Hard | — | 79.8% |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math gpt-oss-120b leads
gpt-oss-120b: 50.6 (#43), Qwen2.5 72B Instruct: 18.1 (#281)
| Benchmark | gpt-oss-120b | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 8.1% |
| Omni-MATH | 68.8% | 33% |
| LMArena Math | 1388 | 1283 |
| MATH Level 5 | — | 63.2% |
Knowledge gpt-oss-120b leads
gpt-oss-120b: 42.3 (#89), Qwen2.5 72B Instruct: 26.9 (#247)
| Benchmark | gpt-oss-120b | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 75.8% | 49.1% |
| MMLU-Pro | 79.5% | 63.1% |
| Confabulations | 15.7% | 19.1% |
| GPQA (HELM) | 68.4% | 42.6% |
| LMArena Expert | 1356 | 1245 |
| Vectara Hallucination Rate | 14.2% | — |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multilingual gpt-oss-120b leads
gpt-oss-120b: 48.2 (#142), Qwen2.5 72B Instruct: 41.2 (#208)
| Benchmark | gpt-oss-120b | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1350 | 1252 |
| LMArena Chinese | 1384 | 1272 |
| LMArena French | 1370 | 1279 |
| LMArena German | 1355 | 1234 |
| LMArena Japanese | 1332 | 1180 |
| LMArena Korean | 1283 | 1188 |
| LMArena Russian | 1343 | 1264 |
| LMArena Spanish | 1388 | 1255 |
Instruction Following gpt-oss-120b leads
gpt-oss-120b: 69.5 (#168), Qwen2.5 72B Instruct: 65.7 (#216)
| Benchmark | gpt-oss-120b | Qwen2.5 72B Instruct |
|---|---|---|
| IFEval | 83.6% | 80.6% |
| LMArena Instruction Following | 1318 | 1254 |
Long Context Qwen2.5 72B Instruct leads
gpt-oss-120b: 31.4 (#273), Qwen2.5 72B Instruct: 38.9 (#183)
| Benchmark | gpt-oss-120b | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1319 | 1282 |
| Fiction.LiveBench | 44.4% | — |
Writing & Preference Too close to call
gpt-oss-120b: 46.7 (#213), Qwen2.5 72B Instruct: 46.9 (#210)
| Benchmark | gpt-oss-120b | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1366 | 1269 |
| LMArena Creative Writing | 1274 | 1221 |
| WildBench | 84.5% | 80.2% |
| LMArena Multi-Turn | 1340 | 1272 |
| Short-Story Creative Writing | 77.1% | — |
| EQ-Bench Creative Writing | 961 | — |
Frequently asked questions
Is gpt-oss-120b better than Qwen2.5 72B Instruct?
gpt-oss-120b is the stronger model overall, scoring 36.0 to 31.5 on the Noometry Index.
Which is cheaper, gpt-oss-120b or Qwen2.5 72B Instruct?
gpt-oss-120b is cheaper. It lists at $0.037 per million input tokens and $0.17 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is gpt-oss-120b or Qwen2.5 72B Instruct better for coding?
They score almost the same on coding (33.2 vs 32.8); test both on your own repository before choosing.
Which has the bigger context window?
Both accept 131K tokens.
How many benchmarks do gpt-oss-120b and Qwen2.5 72B Instruct share?
30 benchmarks have published results for both models. gpt-oss-120b has 48 scored results on Noometry and Qwen2.5 72B Instruct has 43.