Vals Science Index
Vals AI's composite of its biology investigation, mechanism discovery and research workflow evaluations, weighted equally. Run by Vals AI.
As of 24 Oct 2026, GPT-6 Astra leads Vals Science Index on BenchLeader with 65.1%, ahead of GPT-6.1 Sol at 59.6%, across 22 model configurations with a published result.
- Published by
- Vals AI
- Category
- Composite
- Index weight
- Reference only
- Models
- 22
- Data as of
- 24 Oct 2026
- Newest result
- 21 Oct 2026
Vals AI (vals.ai).
- 1GPT-6 Astra (max)65.1%
- 2GPT-6.1 Sol (max)59.6%
- 3Claude Sonnet 5.559.1%
- 4Claude Opus 5.558.6%
- 5Gemini 4 Argon (high)55.4%
- 6Muse Wonder Full (max)48.5%
- 7Claude Opus 547.9%
- 8GPT-6 Sol (max)45.0%
- 9GPT-5.6 Sol (max)41.5%
- 10Muse Wonder (max)39.1%
- 11Claude Haiku 5.538.5%
- 12Grok 4.6 (high)38.1%
- 13Gemini 3.8 Flash (high)35.8%
- 14Grok 4.7 (xhigh)34.8%
- 15Ember-131.8%
22 of 22
| # | ||||
|---|---|---|---|---|
| 1 | 65.1% | 70.4 | 2026-10-21 | |
| 2 | 59.6% | 69.9 | 2026-10-21 | |
| 3 | 59.1% | – | 2026-10-21 | |
| 4 | 58.6% | 62.0 | 2026-10-21 | |
| 5 | 55.4% | 70.8 | 2026-10-21 | |
| 6 | 48.5% | 70.7 | 2026-10-21 | |
| 7 | 47.9% | 65.5 | 2026-10-21 | |
| 8 | 45.0% | 66.4 | 2026-10-21 | |
| 9 | 41.5% | 68.0 | 2026-10-21 | |
| 10 | 39.1% | 67.5 | 2026-10-21 | |
| 11 | 38.5% | – | 2026-10-21 | |
| 12 | 38.1% | 62.9 | 2026-10-21 | |
| 13 | 35.8% | 64.4 | 2026-10-21 | |
| 14 | 34.8% | 61.6 | 2026-10-21 | |
| 15 | Ember-1 | 31.8% | – | 2026-10-21 |
| 16 | 31.7% | 59.7 | 2026-10-21 | |
| 17 | 29.6% | 55.6 | 2026-10-21 | |
| 18 | 29.6% | 60.4 | 2026-10-21 | |
| 19 | 28.4% | 60.0 | 2026-10-21 | |
| 20 | 27.4% | 60.3 | 2026-10-21 | |
| 21 | 25.3% | 59.9 | 2026-10-21 | |
| 22 | New | 22.0% | 51.1 | 2026-10-21 |
More composite benchmarks · All LLM benchmarks
Cite as: BenchLeader, “Vals Science Index leaderboard”, data as of 24 Oct 2026, https://www.benchleader.com/benchmarks/vals_science_index.
Vals Science Index: questions
- What does Vals Science Index measure?
- Vals AI's composite of its biology investigation, mechanism discovery and research workflow evaluations, weighted equally. Run by Vals AI. Scores are reported in percent of tasks solved; higher is better.
- Which AI model leads Vals Science Index?
- GPT-6 Astra leads Vals Science Index with 65.1% as of 24 Oct 2026, ahead of GPT-6.1 Sol at 59.6%.
- How many models have Vals Science Index results?
- 22 model configurations have a Vals Science Index result on BenchLeader, all taken from Vals AI.
- Who runs Vals Science Index and how often is it updated?
- Vals Science Index is published by Vals AI. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does Vals Science Index count toward the BenchLeader Index?
- No. Vals Science Index is shown for reference but left out of the composite index.