Benchmark
Vals ProgramBench
Coding benchmarks compare models on programming tasks. Scores are percentages.
- Models measured
- 64
- Scores recorded
- 65
- Sources
- 1
- IQ weight
- 3.42
Score distribution
- 1.3%
- Mean
- 0%
- Median
- 3.4%
- Std dev
- 0%
- 10th percentile
- 3%
- 90th percentile
- 0%
- Min
This benchmark has a weight of 3.42 for chat models, measured over 64 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 | Grok 4.3· Vals AI | 0% |
| 52 | Grok 4.5· Vals AI | 0% |
| 53 | Hy4 preview· Vals AI | 0% |
| 54 | Inkling FP4· Vals AI | 0% |
| 55 | Kimi K2.6· Vals AI | 0% |
| 56 | Kimi K2.7 Code· Vals AI | 0% |
| 57 | Laguna M.1· Vals AI | 0% |
| 58 | Laguna XS.2· Vals AI | 0% |
| 59 | MiniMax M2.7· Vals AI | 0% |
| 60 | Muse Spark 1.1· Vals AI | 0% |
| 61 | Nemotron 3 Ultra· Vals AI | 0% |
| 62 | Nemotron 3.5 Lightning 30B A3B· Vals AI | 0% |
| 63 | Qwen 3.8 27B· Vals AI | 0% |
| 64 | Qwen3.6 Plus· Vals AI | 0% |
| 65 | Qwen3.8 Max· Vals AI | 0% |
Showing 51–65 of 65 results.