OpenAI: gpt-oss-20b wins on more metrics (5 of 9), but the right pick depends on what you optimize for — see the breakdown below.
| Metric | OpenAI: gpt-oss-20b | Qwen: Qwen3 Coder 480B A35B |
|---|---|---|
| Intelligence Index | 24.5 | 24.8✓ |
| Coding Index | 18.5 | 24.6✓ |
| GPQA Diamond | 69%✓ | 62% |
| Design Arena Elo | 977 | 1203✓ |
| Speed (tokens/sec) | 397✓ | 47 |
| Latency | 223ms✓ | 658ms |
| Input price /M | $0.029✓ | $0.220 |
| Output price /M | $0.140✓ | $1.80 |
| Context window | 131K | 1.0M✓ |
| Capabilities | ReasoningToolsJSON | ToolsJSON |