Benchmark
Vals Science Index
Scores are percentages.
Benchmark sourceScores from Vals AI, LLM Stats data.
- Models measured
- 27
- Scores recorded
- 27
- Sources
- 2
- IQ weight
- 2.15
Score distribution
- 44.4%
- Mean
- 41.5%
- Median
- 14%
- Std dev
- 28.4%
- 10th percentile
- 63.7%
- 90th percentile
- 22%
- Min
This benchmark has a weight of 2.15 for chat models, measured over 27 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 | 68.4% |
| 2 | GPT-6 Astra· Vals AI | 65.1% |
| 3 | Claude Fable 5.2· Vals AI | 63.7% |
| 4 | GPT-6.1 Sol· Vals AI | 59.6% |
| 5 | Claude Sonnet 5.5· Vals AI | 59.1% |
| 6 | Claude Opus 5.5· Vals AI | 58.7% |
| 7 | GPT-6.2 Sol· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 58.5% |
| 8 | Gemini 4 Argon· Vals AI | 55.4% |
| 9 | Muse Wonder Full· Vals AI | 48.5% |
| 10 | Claude Opus 5· Vals AI | 47.9% |
| 11 | GPT-6 Sol· Vals AI | 45% |
| 12 | Kimi K3.1 Pro· Vals AI | 44.4% |
| 13 | GPT-5.6 Sol· Vals AI | 41.5% |
| 14 | Muse Wonder· Vals AI | 39.1% |
| 15 | Claude Haiku 5.5· Vals AI | 38.5% |
| 16 | Grok 4.6· Vals AI | 38.1% |
| 17 | Gemini 3.8 Flash· Vals AI | 35.8% |
| 18 | Grok 4.7· Vals AI | 34.8% |
| 19 | Inkling 2 Small· Vals AI | 32.7% |
| 20 | Ember-1· Vals AI | 31.8% |
| 21 | MiMo V2.6 Flash· Vals AI | 31.7% |
| 22 | DeepSeek V4.1 Flash· Vals AI | 29.7% |
| 23 | Step 5 Preview· Vals AI | 29.6% |
| 24 | GPT-6 Luna· Vals AI | 28.4% |
| 25 | Gemini 3.1 Pro Preview· Vals AI | 27.4% |
| 26 | GPT-5.6 Luna· Vals AI | 25.3% |
| 27 | Beam 1.0 501B A23B· Vals AI | 22% |
Data credits: Data by LLM Stats