Model comparison
gpt-oss-120b vs Qwen3 32B
Qwen3 32B is the stronger model overall, scoring 39.2 to 36.3 on the Noometry Index. gpt-oss-120b costs 17× less per token, which makes it the better buy when Qwen3 32B's lead doesn't matter for your workload.
Last verified . 25 shared benchmarks.
Alibaba (Qwen)Summary
- They share 25 benchmarks with published results for both. gpt-oss-120b scores higher in 4 categories and Qwen3 32B in 5 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where Qwen3 32B leads 32.6 to 12.2.
- The biggest single-benchmark swing is Fiction.LiveBench: 44.4% for gpt-oss-120b and 74.2% for Qwen3 32B.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $0.70 / $2.80 for Qwen3 32B.
Side by side
| gpt-oss-120b | Qwen3 32B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 36.3 | 39.2 |
| Released | 2025-08-05 | 2025-04 |
| Weights | Open | Open |
| Context window | 131K | 131K |
| Max output | 41K | 16K |
| Input $ / M tokens | $0.037 | $0.70 |
| Output $ / M tokens | $0.17 | $2.80 |
| Results tracked | 48 | 26 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3 32B leads
gpt-oss-120b: 33.5 (#256), Qwen3 32B: 37.7 (#190)
| Benchmark | gpt-oss-120b | Qwen3 32B |
|---|---|---|
| Aider Polyglot | 41.8% | 40% |
| SciCode | 36% | 35.4% |
| LMArena Coding | 1380 | 1358 |
| SWE-bench Verified (bash only) | 26% | — |
| WeirdML | 48.2% | — |
| ALE-Bench | 575.62 | — |
| AlgoTune | 1.41 | — |
Agentic & Tool Use Qwen3 32B leads
gpt-oss-120b: 12.2 (#153), Qwen3 32B: 32.6 (#62)
| Benchmark | gpt-oss-120b | Qwen3 32B |
|---|---|---|
| Terminal-Bench | 18.7% | — |
| APEX-Agents | 4.4% | — |
| Berkeley Function Calling Leaderboard | — | 48.7% |
| METR Time Horizons | 56.6% | — |
| Vending-Bench 2 | -21.53 | — |
Reasoning Too close to call
gpt-oss-120b: 20.0 (#245), Qwen3 32B: 20.2 (#241)
| Benchmark | gpt-oss-120b | Qwen3 32B |
|---|---|---|
| Kagi LLM Benchmark | 58.6% | 54.9% |
| CritPt | 1.1% | 0.3% |
| Chess Puzzles | 20% | 5% |
| LMArena Hard Prompts | 1364 | 1334 |
| DTBench | 76.3% | 67.5% |
| LMCA | 22.1% | 17.3% |
| Epoch Capabilities Index | 139.93 | 138.51 |
| SimpleBench | 22.1% | — |
| Mystery Game Puzzles | 2% | — |
| Surface Evolver Bench | 25% | — |
Math gpt-oss-120b leads
gpt-oss-120b: 52.5 (#50), Qwen3 32B: 39.7 (#99)
| Benchmark | gpt-oss-120b | Qwen3 32B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 66.9% |
| LMArena Math | 1389 | 1399 |
| Omni-MATH | 68.8% | — |
Knowledge gpt-oss-120b leads
gpt-oss-120b: 42.4 (#96), Qwen3 32B: 40.0 (#125)
| Benchmark | gpt-oss-120b | Qwen3 32B |
|---|---|---|
| GPQA Diamond | 75.8% | 65.7% |
| Vectara Hallucination Rate | 14.2% | 5.9% |
| LMArena Expert | 1356 | 1362 |
| MMLU-Pro | 79.5% | — |
| Confabulations | 15.7% | — |
| GPQA (HELM) | 68.4% | — |
Multilingual gpt-oss-120b leads
gpt-oss-120b: 48.0 (#147), Qwen3 32B: 45.6 (#167)
| Benchmark | gpt-oss-120b | Qwen3 32B |
|---|---|---|
| LMArena Non-English | 1351 | 1317 |
| LMArena Chinese | 1385 | 1357 |
| LMArena German | 1353 | 1341 |
| LMArena Russian | 1343 | 1311 |
| LMArena French | 1369 | — |
| LMArena Japanese | 1331 | — |
| LMArena Korean | 1282 | — |
| LMArena Spanish | 1389 | — |
Instruction Following Too close to call
gpt-oss-120b: 69.3 (#173), Qwen3 32B: 68.9 (#179)
| Benchmark | gpt-oss-120b | Qwen3 32B |
|---|---|---|
| LMArena Instruction Following | 1318 | 1305 |
| IFEval | 83.6% | — |
Long Context Qwen3 32B leads
gpt-oss-120b: 31.4 (#278), Qwen3 32B: 43.8 (#87)
| Benchmark | gpt-oss-120b | Qwen3 32B |
|---|---|---|
| Fiction.LiveBench | 44.4% | 74.2% |
| LMArena Longer Query | 1319 | 1327 |
Writing & Preference Qwen3 32B leads
gpt-oss-120b: 46.5 (#217), Qwen3 32B: 52.9 (#163)
| Benchmark | gpt-oss-120b | Qwen3 32B |
|---|---|---|
| LMArena Text | 1365 | 1340 |
| LMArena Creative Writing | 1275 | 1297 |
| LMArena Multi-Turn | 1340 | 1331 |
| Short-Story Creative Writing | 77.1% | — |
| EQ-Bench Creative Writing | 961 | — |
| WildBench | 84.5% | — |
Frequently asked questions
Is gpt-oss-120b better than Qwen3 32B?
Qwen3 32B is the stronger model overall, scoring 39.2 to 36.3 on the Noometry Index. gpt-oss-120b costs 17× less per token, which makes it the better buy when Qwen3 32B's lead doesn't matter for your workload.
Which is cheaper, gpt-oss-120b or Qwen3 32B?
gpt-oss-120b is cheaper. It lists at $0.037 per million input tokens and $0.17 per million output tokens; Qwen3 32B lists at $0.70 and $2.80.
Is gpt-oss-120b or Qwen3 32B better for coding?
Qwen3 32B scores higher on coding benchmarks: 37.7 versus 33.5 in the Noometry coding category.
Which has the bigger context window?
Both accept 131K tokens.
How many benchmarks do gpt-oss-120b and Qwen3 32B share?
25 benchmarks have published results for both models. gpt-oss-120b has 48 scored results on Noometry and Qwen3 32B has 26.