Benchmark
Vals SWE-bench Multimodal
Coding benchmarks compare models on programming tasks. Scores are percentages.
- Models measured
- 21
- Scores recorded
- 21
- Sources
- 1
- IQ weight
- 3.60
Score distribution
- 46.6%
- Mean
- 43.5%
- Median
- 9.4%
- Std dev
- 37.5%
- 10th percentile
- 61.3%
- 90th percentile
- 33.1%
- Min
This benchmark has a weight of 3.60 for chat models, measured over 21 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 Opus 5.5· Vals AI | 70.2% |
| 2 | Claude Fable 5.2· Vals AI | 62.3% |
| 3 | Claude Fable 5.1· Vals AI | 61.3% |
| 4 | Claude Sonnet 5.5· Vals AI | 58.3% |
| 5 | Gemini 4 Argon· Vals AI | 56% |
| 6 | Muse Wonder Full· Vals AI | 47.3% |
| 7 | DeepSeek V4.1 Flash· Vals AI | 46.5% |
| 8 | Muse Wonder· Vals AI | 45.6% |
| 9 | Kimi K3· Vals AI | 44.2% |
| 10 | Claude Haiku 5.5· Vals AI | 44% |
| 11 | GPT-6.1 Sol· Vals AI | 43.5% |
| 12 | GPT-6 Astra· Vals AI | 43.3% |
| 13 | Inkling 2 Small· Vals AI | 43.1% |
| 14 | Grok 4.7· Vals AI | 42.9% |
| 15 | Qwen3.8 Max· Vals AI | 42.5% |
| 16 | MiMo V2.6 Pro· Vals AI | 41.3% |
| 17 | Gemini 3.8 Flash· Vals AI | 40.8% |
| 18 | Step 5 Preview· Vals AI | 40.8% |
| 19 | GPT-6 Luna· Vals AI | 37.5% |
| 20 | MiniMax M3· Vals AI | 33.3% |
| 21 | Mistral Large 4· Vals AI | 33.1% |