gpt-oss-20b vs Qwen3-235B-A22B
Benchmarks, API pricing and specs, head to head. Data updated 2026-06-10.
OpenAI · Aug 2025
The small gpt-oss variant that runs on a 16GB consumer GPU or laptop — local reasoning for everyone, Apache 2.0 licensed.
Alibaba (Qwen) · Jul 2025
The most popular open general-purpose Qwen: 235B MoE with hybrid thinking, a staple of the 2025 open-source ecosystem.
The verdict
These two models don't yet share verified results on the benchmarks we track, so judge them on specs, pricing and intended use. On price the gap is dramatic: gpt-oss-20b works out roughly 4.4x cheaper per blended million tokens. Qwen3-235B-A22B also takes 262K of context versus 131K for gpt-oss-20b.
Specs & pricing
| gpt-oss-20b | Qwen3-235B-A22B | |
|---|---|---|
| modhub Index | — | 69.0 |
| Input price / 1M | $0.05 | $0.22 |
| Output price / 1M | $0.2 | $0.88 |
| Context window | 131K | 262K |
| Max output | 131K | 33K |
| Open weights | yes (Apache 2.0) | yes (Apache 2.0) |
| Reasoning model | yes | yes |
| Multimodal input | text | text |
| Knowledge cutoff | Jun 2024 | Apr 2025 |
| Released | Aug 2025 | Jul 2025 |
| Example monthly cost* | $0.8 | $3.52 |
* 10M input + 1.5M output tokens per month at list prices, no caching. Green = better value on that row.
Frequently asked questions
- Which is better, gpt-oss-20b or Qwen3-235B-A22B?
- These two models don't yet share verified results on the benchmarks we track, so judge them on specs, pricing and intended use. On price the gap is dramatic: gpt-oss-20b works out roughly 4.4x cheaper per blended million tokens. Qwen3-235B-A22B also takes 262K of context versus 131K for gpt-oss-20b.
- Which is cheaper, gpt-oss-20b or Qwen3-235B-A22B?
- gpt-oss-20b costs $0.05/$0.2 per million input/output tokens, while Qwen3-235B-A22B costs $0.22/$0.88. For a typical workload of 10M input and 1.5M output tokens per month, that's $0.8 versus $3.52.
- Which model is better for coding, gpt-oss-20b or Qwen3-235B-A22B?
- We don't yet track SWE-bench Verified results for both models; check their individual pages for coding-related scores.