Benchmark
QwenSWEBench
Coding benchmarks compare models on programming tasks. Scores are percentages. Higher is better.
- Models measured
- 6
- Scores recorded
- 8
- Sources
- 2
- IQ weight
- 1.02
Score distribution
- 78.3%
- Mean
- 80.7%
- Median
- 6.7%
- Std dev
- 73.5%
- 10th percentile
- 84%
- 90th percentile
- 63.4%
- Min
This benchmark has a weight of 1.02 for chat models, measured over 6 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5Qwen内部基准:Claude Code · avg@3 · 8h · 256K· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 86.3% |
| 2 | Claude Opus 4.8Qwen内部基准:Claude Code · avg@3 · 8h · 256K· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 84% |
| 3 | Qwen3.8 Max· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 80.7% |
| 4 | Qwen3.8 MaxQwen内部基准:Claude Code · avg@3 · 8h · 256K· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 80.7% |
| 5 | Qwen 3.8 27B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 79% |
| 6 | Qwen 3.8 27BQwen3.8-27B model card · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 79% |
| 7 | GPT-5.6 SolQwen内部基准:Claude Code · avg@3 · 8h · 256K· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 73.5% |
| 8 | Qwen3.7 MaxQwen内部基准:Claude Code · avg@3 · 8h · 256K· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 63.4% |
Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Data by LLM Stats