Model comparison
gpt-oss-20b vs Qwen3.8 27B
Qwen3.8 27B is the stronger model overall, scoring 46.0 to 32.5 on the Noometry Index. gpt-oss-20b costs 31× less per token, which makes it the better buy when Qwen3.8 27B's lead doesn't matter for your workload.
Last verified . 22 shared benchmarks.
Alibaba (Qwen)Summary
- They share 22 benchmarks with published results for both. gpt-oss-20b scores higher in 1 category and Qwen3.8 27B in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Qwen3.8 27B leads 65.8 to 35.5.
- The biggest single-benchmark swing is LMCA: 14.5% for gpt-oss-20b and 41.4% for Qwen3.8 27B.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $0.99 / $1.49 for Qwen3.8 27B.
- Qwen3.8 27B accepts more context: 262K tokens versus 131K.
Side by side
| gpt-oss-20b | Qwen3.8 27B | |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 32.5 | 46.0 |
| Released | 2025-08-05 | 2026-08-14 |
| Weights | Open | Open |
| Context window | 131K | 262K |
| Max output | 16K | 33K |
| Input $ / M tokens | $0.018 | $0.99 |
| Output $ / M tokens | $0.09 | $1.49 |
| Results tracked | 34 | 31 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 27B leads
gpt-oss-20b: 37.6 (#192), Qwen3.8 27B: 50.5 (#44)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| SciCode | 34.4% | 46.6% |
| LMArena Coding | 1306 | 1482 |
| LMArena WebDev | — | 1593 |
| WeirdML | 40.9% | — |
| ALE-Bench | 566.05 | — |
Agentic & Tool Use Qwen3.8 27B leads
gpt-oss-20b: 9.3 (#154), Qwen3.8 27B: 32.9 (#57)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| Terminal-Bench | 3.4% | — |
| APEX-Agents | — | 47.5% |
Reasoning Qwen3.8 27B leads
gpt-oss-20b: 19.3 (#261), Qwen3.8 27B: 41.0 (#54)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| CritPt | 1.4% | 5.4% |
| LMArena Hard Prompts | 1274 | 1460 |
| DTBench | 68% | 88% |
| LMCA | 14.5% | 41.4% |
| Epoch Capabilities Index | 137.82 | 149.38 |
| ARC-AGI-2 | — | 42.4% |
| Kagi LLM Benchmark | 53.2% | — |
| NYT Connections (extended) | — | 54.5% |
| ARC-AGI-1 | — | 87.5% |
| Chess Puzzles | 4% | — |
| Surface Evolver Bench | — | 45% |
Math gpt-oss-20b leads
gpt-oss-20b: 39.4 (#103), Qwen3.8 27B: 37.1 (#161)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| LMArena Math | 1317 | 1456 |
| OTIS Mock AIME 2024-2025 | 65.3% | — |
| ProofBench | — | 16% |
| Omni-MATH | 56.5% | — |
Knowledge Qwen3.8 27B leads
gpt-oss-20b: 34.6 (#195), Qwen3.8 27B: 41.6 (#109)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| LMArena Expert | 1258 | 1482 |
| GPQA Diamond | 60.8% | — |
| MMLU-Pro | 74% | — |
| GPQA (HELM) | 59.4% | — |
Multimodal Not comparable
gpt-oss-20b: —, Qwen3.8 27B: 41.3 (#37)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| LMArena Vision | — | 1271 |
Multilingual Qwen3.8 27B leads
gpt-oss-20b: 42.2 (#197), Qwen3.8 27B: 53.7 (#60)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| LMArena Non-English | 1268 | 1430 |
| LMArena Chinese | 1314 | 1504 |
| LMArena German | 1255 | 1438 |
| LMArena Japanese | 1244 | 1384 |
| LMArena Korean | 1236 | 1393 |
| LMArena Russian | 1278 | 1415 |
| LMArena Spanish | 1267 | 1448 |
| LMArena French | — | 1465 |
Instruction Following Qwen3.8 27B leads
gpt-oss-20b: 61.8 (#240), Qwen3.8 27B: 75.8 (#53)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| LMArena Instruction Following | 1236 | 1439 |
| IFEval | 73.2% | — |
Long Context Qwen3.8 27B leads
gpt-oss-20b: 37.9 (#209), Qwen3.8 27B: 44.3 (#70)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| LMArena Longer Query | 1250 | 1450 |
Writing & Preference Qwen3.8 27B leads
gpt-oss-20b: 35.5 (#265), Qwen3.8 27B: 65.8 (#43)
| Benchmark | gpt-oss-20b | Qwen3.8 27B |
|---|---|---|
| LMArena Text | 1287 | 1441 |
| LMArena Creative Writing | 1201 | 1384 |
| EQ-Bench Creative Writing | 666 | 1671 |
| LMArena Multi-Turn | 1268 | 1441 |
| WildBench | 73.7% | — |
Frequently asked questions
Is gpt-oss-20b better than Qwen3.8 27B?
Qwen3.8 27B is the stronger model overall, scoring 46.0 to 32.5 on the Noometry Index. gpt-oss-20b costs 31× less per token, which makes it the better buy when Qwen3.8 27B's lead doesn't matter for your workload.
Which is cheaper, gpt-oss-20b or Qwen3.8 27B?
gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; Qwen3.8 27B lists at $0.99 and $1.49.
Is gpt-oss-20b or Qwen3.8 27B better for coding?
Qwen3.8 27B scores higher on coding benchmarks: 50.5 versus 37.6 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 27B does, with 262K tokens against 131K.
How many benchmarks do gpt-oss-20b and Qwen3.8 27B share?
22 benchmarks have published results for both models. gpt-oss-20b has 34 scored results on Noometry and Qwen3.8 27B has 31.