Benchmark
Vals Public Benefits Bench v1
Scores are percentages.
Benchmark sourceScores from Vals AI.
- Models measured
- 13
- Scores recorded
- 13
- Sources
- 1
- IQ weight
- 1.39
Score distribution
- 56.3%
- Mean
- 57.6%
- Median
- 6.6%
- Std dev
- 49.5%
- 10th percentile
- 62.1%
- 90th percentile
- 44.3%
- Min
This benchmark has a weight of 1.39 for chat models, measured over 13 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 Fable 5· Vals AI | 71.7% |
| 2 | Claude Opus 4.8· Vals AI | 62.1% |
| 3 | MiniMax M3· Vals AI | 60.7% |
| 4 | Claude Sonnet 4.6· Vals AI | 58.5% |
| 5 | Gemini 3.5 Flash· Vals AI | 58% |
| 6 | GLM 5.1· Vals AI | 57.9% |
| 7 | DeepSeek V4 Pro· Vals AI | 57.6% |
| 8 | GPT 5.5· Vals AI | 57.2% |
| 9 | Kimi K2.6· Vals AI | 53.4% |
| 10 | Gemini 3.1 Pro Preview· Vals AI | 50.9% |
| 11 | Grok 4.3· Vals AI | 50.1% |
| 12 | Claude Haiku 4.5· Vals AI | 49.5% |
| 13 | Grok 4.1 Fast· Vals AI | 44.3% |