Benchmark
SWE-rebench 2026-06
Coding benchmarks compare models on programming tasks. Scores are percentages.
- Models measured
- 8
- Scores recorded
- 8
- Sources
- 1
- IQ weight
- 0.94
Score distribution
- 35.4%
- Mean
- 35.2%
- Median
- 15.3%
- Std dev
- 20%
- 10th percentile
- 55.7%
- 90th percentile
- 13.8%
- Min
This benchmark has a weight of 0.94 for chat models, measured over 8 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | |
|---|---|---|
| 1 | GLM 5.2tools, high· SWE-rebench | 58.1% |
| 2 | Grok 4.5tools, high· SWE-rebench | 55.7% |
| 3 | MiniMax M3tools· SWE-rebench | 43.8% |
| 4 | DeepSeek V4 Protools, high· SWE-rebench | 35.2% |
| 5 | MiMo V2.5 Protools· SWE-rebench | 33.8% |
| 6 | Qwen3.6 27Btools· SWE-rebench | 22.9% |
| 7 | Qwen3.6 35B A3Btools· SWE-rebench | 20% |
| 8 | Qwen3.5 35B A3Btools· SWE-rebench | 13.8% |
Tokens per task, input and output together, as measured by the source. Comparable within this benchmark only.