Model comparison
gpt-oss-20b vs Qwen2.5 72B Instruct
gpt-oss-20b and Qwen2.5 72B Instruct score almost the same on the Noometry Index (32.1 vs 31.5), so choose on price, context window or the category you care about most.
Last verified . 27 shared benchmarks.
Alibaba (Qwen)Summary
- They share 27 benchmarks with published results for both. gpt-oss-20b scores higher in 4 categories and Qwen2.5 72B Instruct in 5 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where gpt-oss-20b leads 37.6 to 18.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 65.3% for gpt-oss-20b and 8.1% for Qwen2.5 72B Instruct.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $1.40 / $5.60 for Qwen2.5 72B Instruct.
Side by side
| gpt-oss-20b | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 32.1 | 31.5 |
| Released | 2025-08-05 | 2024-09 |
| Weights | Open | Open |
| Context window | 131K | 131K |
| Max output | 16K | 8K |
| Input $ / M tokens | $0.018 | $1.40 |
| Output $ / M tokens | $0.09 | $5.60 |
| Results tracked | 34 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding gpt-oss-20b leads
gpt-oss-20b: 37.1 (#185), Qwen2.5 72B Instruct: 32.8 (#256)
| Benchmark | gpt-oss-20b | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | 40.9% | 16% |
| LMArena Coding | 1307 | 1293 |
| SciCode | 34.4% | — |
| BigCodeBench Instruct | — | 45.8% |
| BigCodeBench Complete | — | 55.9% |
| ALE-Bench | 566.05 | — |
Agentic & Tool Use Qwen2.5 72B Instruct leads
gpt-oss-20b: 9.2 (#146), Qwen2.5 72B Instruct: 22.0 (#124)
| Benchmark | gpt-oss-20b | Qwen2.5 72B Instruct |
|---|---|---|
| Terminal-Bench | 3.4% | — |
| TheAgentCompany | — | 5.7% |
| BALROG | — | 16.2% |
| METR Time Horizons | — | 35.8% |
Reasoning Qwen2.5 72B Instruct leads
gpt-oss-20b: 18.8 (#251), Qwen2.5 72B Instruct: 21.4 (#197)
| Benchmark | gpt-oss-20b | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1274 | 1271 |
| DTBench | 68% | 62.9% |
| LMCA | 14.5% | 13.4% |
| Epoch Capabilities Index | 137.81 | 129.01 |
| Kagi LLM Benchmark | 53.2% | — |
| CritPt | 1.4% | — |
| Chess Puzzles | 4% | — |
| BIG-Bench Hard | — | 79.8% |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math gpt-oss-20b leads
gpt-oss-20b: 37.6 (#93), Qwen2.5 72B Instruct: 18.1 (#281)
| Benchmark | gpt-oss-20b | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 65.3% | 8.1% |
| Omni-MATH | 56.5% | 33% |
| LMArena Math | 1317 | 1283 |
| MATH Level 5 | — | 63.2% |
Knowledge gpt-oss-20b leads
gpt-oss-20b: 34.6 (#186), Qwen2.5 72B Instruct: 26.9 (#247)
| Benchmark | gpt-oss-20b | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 60.8% | 49.1% |
| MMLU-Pro | 74% | 63.1% |
| GPQA (HELM) | 59.4% | 42.6% |
| LMArena Expert | 1258 | 1245 |
| Confabulations | — | 19.1% |
| ARC (AI2) Challenge | — | 94.5% |
| MMLU | — | 85.3% |
| TriviaQA | — | 71.9% |
Multilingual gpt-oss-20b leads
gpt-oss-20b: 42.4 (#192), Qwen2.5 72B Instruct: 41.2 (#208)
| Benchmark | gpt-oss-20b | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1269 | 1252 |
| LMArena Chinese | 1311 | 1272 |
| LMArena German | 1249 | 1234 |
| LMArena Japanese | 1245 | 1180 |
| LMArena Korean | 1239 | 1188 |
| LMArena Russian | 1279 | 1264 |
| LMArena Spanish | 1267 | 1255 |
| LMArena French | — | 1279 |
Instruction Following Qwen2.5 72B Instruct leads
gpt-oss-20b: 62.0 (#235), Qwen2.5 72B Instruct: 65.7 (#216)
| Benchmark | gpt-oss-20b | Qwen2.5 72B Instruct |
|---|---|---|
| IFEval | 73.2% | 80.6% |
| LMArena Instruction Following | 1236 | 1254 |
Long Context Too close to call
gpt-oss-20b: 37.9 (#203), Qwen2.5 72B Instruct: 38.9 (#183)
| Benchmark | gpt-oss-20b | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1251 | 1282 |
Writing & Preference Qwen2.5 72B Instruct leads
gpt-oss-20b: 35.8 (#260), Qwen2.5 72B Instruct: 46.9 (#210)
| Benchmark | gpt-oss-20b | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1288 | 1269 |
| LMArena Creative Writing | 1201 | 1221 |
| WildBench | 73.7% | 80.2% |
| LMArena Multi-Turn | 1268 | 1272 |
| EQ-Bench Creative Writing | 666 | — |
Frequently asked questions
Is gpt-oss-20b better than Qwen2.5 72B Instruct?
gpt-oss-20b and Qwen2.5 72B Instruct score almost the same on the Noometry Index (32.1 vs 31.5), so choose on price, context window or the category you care about most.
Which is cheaper, gpt-oss-20b or Qwen2.5 72B Instruct?
gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; Qwen2.5 72B Instruct lists at $1.40 and $5.60.
Is gpt-oss-20b or Qwen2.5 72B Instruct better for coding?
gpt-oss-20b scores higher on coding benchmarks: 37.1 versus 32.8 in the Noometry coding category.
Which has the bigger context window?
Both accept 131K tokens.
How many benchmarks do gpt-oss-20b and Qwen2.5 72B Instruct share?
27 benchmarks have published results for both models. gpt-oss-20b has 34 scored results on Noometry and Qwen2.5 72B Instruct has 43.