Benchmark
Vals SWE-bench
Coding benchmarks compare models on programming tasks. Scores are percentages.
- Models measured
- 83
- Scores recorded
- 83
- Sources
- 1
- IQ weight
- 3.33
Score distribution
- 73.7%
- Mean
- 75%
- Median
- 14.7%
- Std dev
- 57.6%
- 10th percentile
- 93%
- 90th percentile
- 7.8%
- Min
This benchmark has a weight of 3.33 for chat models, measured over 83 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 |
|---|---|---|
| 51 | GPT-5.2 Codex· Vals AI | 72.4% |
| 52 | Grok 4.20· Vals AI | 72.2% |
| 53 | GLM 5thinking· Vals AI | 71.4% |
| 54 | Grok 4.3· Vals AI | 71.4% |
| 55 | Qwen3.5 Plusthinking· Vals AI | 71.2% |
| 56 | MiMo V2.5· Vals AI | 71% |
| 57 | Claude Sonnet 4.5· Vals AI | 70% |
| 58 | Kimi K2.5thinking· Vals AI | 70% |
| 59 | Qwen3.6 27B· Vals AI | 70% |
| 60 | GPT 5.1· Vals AI | 69.8% |
| 61 | GPT 5.4 Nano· Vals AI | 69.8% |
| 62 | GLM 4.7· Vals AI | 69.4% |
| 63 | GPT 5· Vals AI | 69% |
| 64 | Nemotron 3 Ultra· Vals AI | 69% |
| 65 | Qwen3.7 Max· Vals AI | 68.8% |
| 66 | DeepSeek V3.2· Vals AI | 67.6% |
| 67 | Claude Haiku 4.5thinking· Vals AI | 66.6% |
| 68 | Mistral Medium 3.5· Vals AI | 66.4% |
| 69 | Qwen3.5 Flash· Vals AI | 64.4% |
| 70 | Devstral 2 123b Instruct 2512· Vals AI | 62.8% |
| 71 | Gemini 3.1 Flash Lite· Vals AI | 62.8% |
| 72 | GPT-5 mini· Vals AI | 60.8% |
| 73 | Kimi K2 Thinking· Vals AI | 60.2% |
| 74 | Grok 4· Vals AI | 57.8% |
| 75 | Laguna M.1· Vals AI | 57.6% |
| 76 | Laguna XS.2· Vals AI | 55.2% |
| 77 | Gemini 2.5 Pro· Vals AI | 54.4% |
| 78 | Nemotron 3.5 Lightning 30B A3B· Vals AI | 52.8% |
| 79 | Grok 4 Fast· Vals AI | 45.4% |
| 80 | Grok 4.1 Fast· Vals AI | 41.4% |
| 81 | Mistral Large 3· Vals AI | 41.4% |
| 82 | GPT OSS 120B· Vals AI | 33.6% |
| 83 | Command A 03 2025· Vals AI | 7.8% |
Showing 51–83 of 83 results.