Model comparison
gpt-oss-20b vs Qwen2-72B
gpt-oss-20b is the stronger model overall, scoring 32.5 to 30.0 on the Noometry Index.
Last verified . 19 shared benchmarks.
Alibaba (Qwen)Summary
- They share 19 benchmarks with published results for both. gpt-oss-20b scores higher in 6 categories and Qwen2-72B in 3 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where gpt-oss-20b leads 34.6 to 21.2.
- The biggest single-benchmark swing is WeirdML: 40.9% for gpt-oss-20b and 11.3% for Qwen2-72B.
Side by side
| gpt-oss-20b | Qwen2-72B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 32.5 | 30.0 |
| Released | 2025-08-05 | 2024-06-07 |
| Weights | Open | Open |
| Context window | 131K | — |
| Max output | 16K | — |
| Input $ / M tokens | $0.018 | — |
| Output $ / M tokens | $0.09 | — |
| Results tracked | 34 | 26 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding gpt-oss-20b leads
gpt-oss-20b: 37.6 (#192), Qwen2-72B: 29.1 (#310)
| Benchmark | gpt-oss-20b | Qwen2-72B |
|---|---|---|
| WeirdML | 40.9% | 11.3% |
| LMArena Coding | 1306 | 1196 |
| SciCode | 34.4% | — |
| BigCodeBench Instruct | — | 38.5% |
| BigCodeBench Complete | — | 54% |
| ALE-Bench | 566.05 | — |
Agentic & Tool Use Qwen2-72B leads
gpt-oss-20b: 9.3 (#154), Qwen2-72B: 17.0 (#146)
| Benchmark | gpt-oss-20b | Qwen2-72B |
|---|---|---|
| Terminal-Bench | 3.4% | — |
| TheAgentCompany | — | 1.1% |
| METR Time Horizons | — | 29.9% |
Reasoning Qwen2-72B leads
gpt-oss-20b: 19.3 (#261), Qwen2-72B: 23.2 (#181)
| Benchmark | gpt-oss-20b | Qwen2-72B |
|---|---|---|
| LMArena Hard Prompts | 1274 | 1191 |
| Epoch Capabilities Index | 137.82 | 125.28 |
| Kagi LLM Benchmark | 53.2% | — |
| CritPt | 1.4% | — |
| Chess Puzzles | 4% | — |
| DTBench | 68% | — |
| LMCA | 14.5% | — |
Math gpt-oss-20b leads
gpt-oss-20b: 39.4 (#103), Qwen2-72B: 30.2 (#236)
| Benchmark | gpt-oss-20b | Qwen2-72B |
|---|---|---|
| LMArena Math | 1317 | 1235 |
| OTIS Mock AIME 2024-2025 | 65.3% | — |
| Omni-MATH | 56.5% | — |
| MATH Level 5 | — | 39.1% |
Knowledge gpt-oss-20b leads
gpt-oss-20b: 34.6 (#195), Qwen2-72B: 21.2 (#275)
| Benchmark | gpt-oss-20b | Qwen2-72B |
|---|---|---|
| GPQA Diamond | 60.8% | 40.8% |
| LMArena Expert | 1258 | 1171 |
| MMLU-Pro | 74% | — |
| GPQA (HELM) | 59.4% | — |
| MMLU | — | 82.4% |
Multilingual gpt-oss-20b leads
gpt-oss-20b: 42.2 (#197), Qwen2-72B: 35.9 (#244)
| Benchmark | gpt-oss-20b | Qwen2-72B |
|---|---|---|
| LMArena Non-English | 1268 | 1176 |
| LMArena Chinese | 1314 | 1240 |
| LMArena German | 1255 | 1151 |
| LMArena Japanese | 1244 | 1111 |
| LMArena Korean | 1236 | 1083 |
| LMArena Russian | 1278 | 1169 |
| LMArena Spanish | 1267 | 1169 |
| LMArena French | — | 1170 |
Instruction Following Too close to call
gpt-oss-20b: 61.8 (#240), Qwen2-72B: 61.7 (#241)
| Benchmark | gpt-oss-20b | Qwen2-72B |
|---|---|---|
| LMArena Instruction Following | 1236 | 1181 |
| IFEval | 73.2% | — |
Long Context gpt-oss-20b leads
gpt-oss-20b: 37.9 (#209), Qwen2-72B: 36.1 (#235)
| Benchmark | gpt-oss-20b | Qwen2-72B |
|---|---|---|
| LMArena Longer Query | 1250 | 1192 |
Writing & Preference Qwen2-72B leads
gpt-oss-20b: 35.5 (#265), Qwen2-72B: 40.8 (#241)
| Benchmark | gpt-oss-20b | Qwen2-72B |
|---|---|---|
| LMArena Text | 1287 | 1203 |
| LMArena Creative Writing | 1201 | 1181 |
| LMArena Multi-Turn | 1268 | 1196 |
| EQ-Bench Creative Writing | 666 | — |
| WildBench | 73.7% | — |
Frequently asked questions
Is gpt-oss-20b better than Qwen2-72B?
gpt-oss-20b is the stronger model overall, scoring 32.5 to 30.0 on the Noometry Index.
Is gpt-oss-20b or Qwen2-72B better for coding?
gpt-oss-20b scores higher on coding benchmarks: 37.6 versus 29.1 in the Noometry coding category.
How many benchmarks do gpt-oss-20b and Qwen2-72B share?
19 benchmarks have published results for both models. gpt-oss-20b has 34 scored results on Noometry and Qwen2-72B has 26.