Model comparison
gpt-oss-120b vs Qwen3.5 27B
Qwen3.5 27B is the stronger model overall, scoring 41.9 to 36.3 on the Noometry Index. gpt-oss-120b costs 12× less per token, which makes it the better buy when Qwen3.5 27B's lead doesn't matter for your workload.
Last verified . 23 shared benchmarks.
Alibaba (Qwen)Summary
- They share 23 benchmarks with published results for both. gpt-oss-120b scores higher in 2 categories and Qwen3.5 27B in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where gpt-oss-120b leads 52.5 to 38.8.
- The biggest single-benchmark swing is LMCA: 22.1% for gpt-oss-120b and 34% for Qwen3.5 27B.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $0.30 / $2.40 for Qwen3.5 27B.
- Qwen3.5 27B accepts more context: 262K tokens versus 131K.
Side by side
| gpt-oss-120b | Qwen3.5 27B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 36.3 | 41.9 |
| Released | 2025-08-05 | 2026-02-23 |
| Weights | Open | Open |
| Context window | 131K | 262K |
| Max output | 41K | 66K |
| Input $ / M tokens | $0.037 | $0.30 |
| Output $ / M tokens | $0.17 | $2.40 |
| Results tracked | 48 | 28 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.5 27B leads
gpt-oss-120b: 33.5 (#256), Qwen3.5 27B: 38.9 (#168)
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| WeirdML | 48.2% | 39.5% |
| LMArena Coding | 1380 | 1427 |
| ALE-Bench | 575.62 | 349.45 |
| SWE-bench Verified (bash only) | 26% | — |
| Aider Polyglot | 41.8% | — |
| LMArena WebDev | — | 1358 |
| SciCode | 36% | — |
| AlgoTune | 1.41 | — |
Agentic & Tool Use Not comparable
gpt-oss-120b: 12.2 (#153), Qwen3.5 27B: —
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| Vending-Bench 2 | -21.53 | 201.98 |
| Terminal-Bench | 18.7% | — |
| APEX-Agents | 4.4% | — |
| METR Time Horizons | 56.6% | — |
Reasoning Qwen3.5 27B leads
gpt-oss-120b: 20.0 (#245), Qwen3.5 27B: 27.5 (#117)
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| LMArena Hard Prompts | 1364 | 1414 |
| DTBench | 76.3% | 82.4% |
| LMCA | 22.1% | 34% |
| SimpleBench | 22.1% | — |
| Kagi LLM Benchmark | 58.6% | — |
| NYT Connections (extended) | — | 47.9% |
| CritPt | 1.1% | — |
| Chess Puzzles | 20% | — |
| Thematic Generalization | — | 45.5% |
| Mystery Game Puzzles | 2% | — |
| Surface Evolver Bench | 25% | — |
| Epoch Capabilities Index | 139.93 | — |
Math gpt-oss-120b leads
gpt-oss-120b: 52.5 (#50), Qwen3.5 27B: 38.8 (#127)
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| LMArena Math | 1389 | 1429 |
| MathArena Final-Answer Competitions | — | 56.7% |
| OTIS Mock AIME 2024-2025 | 88.9% | — |
| Omni-MATH | 68.8% | — |
Knowledge gpt-oss-120b leads
gpt-oss-120b: 42.4 (#96), Qwen3.5 27B: 38.0 (#150)
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| Vectara Hallucination Rate | 14.2% | 12.1% |
| LMArena Expert | 1356 | 1428 |
| GPQA Diamond | 75.8% | — |
| MMLU-Pro | 79.5% | — |
| Confabulations | 15.7% | — |
| GPQA (HELM) | 68.4% | — |
Multimodal Not comparable
gpt-oss-120b: —, Qwen3.5 27B: 39.4 (#59)
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| LMArena Vision | — | 1241 |
Multilingual Qwen3.5 27B leads
gpt-oss-120b: 48.0 (#147), Qwen3.5 27B: 50.8 (#115)
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| LMArena Non-English | 1351 | 1390 |
| LMArena Chinese | 1385 | 1478 |
| LMArena French | 1369 | 1410 |
| LMArena German | 1353 | 1393 |
| LMArena Japanese | 1331 | 1345 |
| LMArena Korean | 1282 | 1358 |
| LMArena Russian | 1343 | 1390 |
| LMArena Spanish | 1389 | 1407 |
Instruction Following Qwen3.5 27B leads
gpt-oss-120b: 69.3 (#173), Qwen3.5 27B: 73.5 (#119)
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| LMArena Instruction Following | 1318 | 1393 |
| IFEval | 83.6% | — |
Long Context Qwen3.5 27B leads
gpt-oss-120b: 31.4 (#278), Qwen3.5 27B: 43.1 (#106)
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| LMArena Longer Query | 1319 | 1413 |
| Fiction.LiveBench | 44.4% | — |
Writing & Preference Qwen3.5 27B leads
gpt-oss-120b: 46.5 (#217), Qwen3.5 27B: 59.3 (#111)
| Benchmark | gpt-oss-120b | Qwen3.5 27B |
|---|---|---|
| LMArena Text | 1365 | 1409 |
| LMArena Creative Writing | 1275 | 1362 |
| LMArena Multi-Turn | 1340 | 1410 |
| Short-Story Creative Writing | 77.1% | — |
| EQ-Bench Creative Writing | 961 | — |
| WildBench | 84.5% | — |
Frequently asked questions
Is gpt-oss-120b better than Qwen3.5 27B?
Qwen3.5 27B is the stronger model overall, scoring 41.9 to 36.3 on the Noometry Index. gpt-oss-120b costs 12× less per token, which makes it the better buy when Qwen3.5 27B's lead doesn't matter for your workload.
Which is cheaper, gpt-oss-120b or Qwen3.5 27B?
gpt-oss-120b is cheaper. It lists at $0.037 per million input tokens and $0.17 per million output tokens; Qwen3.5 27B lists at $0.30 and $2.40.
Is gpt-oss-120b or Qwen3.5 27B better for coding?
Qwen3.5 27B scores higher on coding benchmarks: 38.9 versus 33.5 in the Noometry coding category.
Which has the bigger context window?
Qwen3.5 27B does, with 262K tokens against 131K.
How many benchmarks do gpt-oss-120b and Qwen3.5 27B share?
23 benchmarks have published results for both models. gpt-oss-120b has 48 scored results on Noometry and Qwen3.5 27B has 28.