Vals AI Science Index: leaderboard
Vals AI Science Index v1, launched October16 2026: equal-weight mean of BioMysteryBench, MysteryMechanism and Terminal-Bench Science scores. Measures agentic scientific discovery and execution across biology, hidden mathematical mechanisms and scientific terminal tasks. Native0–100 index, higher is better; component overlap does not create three independent new experiments. The source labels its question set proprietary. https://www.vals.ai/benchmarks/science_index
Metric: Science Index (%). Source: www.vals.ai. Saturation forecast: Around May 2027. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra | 65.09 | #5 |
| 2 | GPT-6.1 Sol | 59.63 | #8 |
| 3 | Claude Sonnet 5.5 | 59.12 | #10 |
| 4 | Claude Opus 5.5 | 58.65 | #3 |
| 5 | Gemini 4 Argon | 55.36 | #7 |
| 6 | Muse Wonder Full | 48.54 | #11 |
| 7 | Claude Opus 5 | 47.93 | #12 |
| 8 | GPT-6 Sol | 45 | #14 |
| 9 | GPT-5.6 Sol | 41.48 | #18 |
| 10 | Muse Wonder | 39.12 | #16 |
| 11 | Claude Haiku 5.5 | 38.5 | #34 |
| 12 | Grok 4.6 | 38.09 | #25 |
| 13 | Gemini 3.8 Flash | 35.76 | #21 |
| 14 | Grok 4.7 | 34.83 | #26 |
| 15 | Ember-1 | 31.76 | #43 |
No result here: #9 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=vals-ai-science-index · How It Works · Data refreshed daily, snapshot 2026-10-17.