Benchmark
Vals SkillsBench
Coding benchmarks compare models on programming tasks. Scores are percentages.
- Models measured
- 33
- Scores recorded
- 33
- Sources
- 1
- IQ weight
- 2.21
Score distribution
- 51.1%
- Mean
- 52.7%
- Median
- 11.1%
- Std dev
- 38.1%
- 10th percentile
- 62.2%
- 90th percentile
- 18.1%
- Min
This benchmark has a weight of 2.21 for chat models, measured over 33 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 | DeepSeek V4.1 Flash· Vals AI | 69.8% |
| 2 | Grok 4.5· Vals AI | 66% |
| 3 | Gemini 3.7 Flash· Vals AI | 65.9% |
| 4 | GPT 5.5· Vals AI | 62.2% |
| 5 | Claude Fable 5.1· Vals AI | 61.6% |
| 6 | GPT-5.6 Luna· Vals AI | 60.4% |
| 7 | Claude Opus 5· Vals AI | 60.4% |
| 8 | Claude Opus 4.8· Vals AI | 59.2% |
| 9 | Muse Spark 1.1· Vals AI | 59.2% |
| 10 | GPT-5.6 Terra· Vals AI | 58.9% |
| 11 | Gemini 3.8 Flash· Vals AI | 58% |
| 12 | Grok 4.6· Vals AI | 55.8% |
| 13 | Qwen3.7 Plus· Vals AI | 54.3% |
| 14 | GPT-5.6 Sol· Vals AI | 54.1% |
| 15 | DeepSeek V4 Pro 0813· Vals AI | 53.8% |
| 16 | Muse Spark 1.2· Vals AI | 53% |
| 17 | Gemini 3.5 Flash· Vals AI | 52.7% |
| 18 | GPT 5.4· Vals AI | 51.7% |
| 19 | MiniMax M3· Vals AI | 51.5% |
| 20 | DeepSeek V4 Pro· Vals AI | 51.3% |
| 21 | DeepSeek V4 Flash 0731· Vals AI | 50.7% |
| 22 | Kimi K2.7 Code· Vals AI | 50% |
| 23 | Claude Sonnet 4.6· Vals AI | 49.1% |
| 24 | GLM 5.3· Vals AI | 47.5% |
| 25 | Claude Sonnet 5· Vals AI | 46.5% |
| 26 | GLM 5.2· Vals AI | 45.1% |
| 27 | Qwen3.8 Max· Vals AI | 42% |
| 28 | Grok 4.3· Vals AI | 40.6% |
| 29 | GLM 5.3 Flash· Vals AI | 40.2% |
| 30 | Qwen 3.8 27B· Vals AI | 38.1% |
| 31 | Inkling Small· Vals AI | 33.6% |
| 32 | Inkling FP4· Vals AI | 26.1% |
| 33 | Mercury 2.5· Vals AI | 18.1% |