Benchmark
Vals Robo-bench
Science benchmarks measure how well models answer science questions. Scores are percentages.
- Models measured
- 11
- Scores recorded
- 11
- Sources
- 1
- IQ weight
- 1.05
Score distribution
- 11.5%
- Mean
- 5.8%
- Median
- 11.2%
- Std dev
- 0.417%
- 10th percentile
- 28.3%
- 90th percentile
- 0.208%
- Min
This benchmark has a weight of 1.05 for chat models, measured over 11 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 | GPT-6.2 Astra· Vals AI | 32.7% |
| 2 | GPT-6.1 Sol· Vals AI | 28.3% |
| 3 | GPT-6 Astra· Vals AI | 23.3% |
| 4 | Claude Opus 5.5· Vals AI | 14.6% |
| 5 | Muse Wonder Full· Vals AI | 10.4% |
| 6 | Claude Fable 5.1· Vals AI | 5.8% |
| 7 | Qwen3.8 Max· Vals AI | 5.2% |
| 8 | DeepSeek V4.1 Flash 1017· Vals AI | 4.2% |
| 9 | Gemini 4 Argon· Vals AI | 0.833% |
| 10 | Kimi K3.1 Pro· Vals AI | 0.417% |
| 11 | Grok 4.7· Vals AI | 0.208% |