# Vals Science Index v1 (Vals Science Index)

> Vals AI composite of agentic science work: investigating biology datasets, recovering hidden mechanisms through experiments, and running research workflows in a terminal.

Canonical page: https://benchlm.ai/benchmarks/vals-science-index

- Category: [Agentic](/agentic)
- Last updated: October 25, 2026

Evaluation results published by [Vals AI](https://www.vals.ai/benchmarks/science_index). Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.

## About Vals Science Index

- Year: 2026
- Tasks: 3 science benchmarks
- Format: Mean accuracy across three benchmarks
- Difficulty: Agentic biology, mechanism discovery and research workflows
- Paper: [Vals Science Index leaderboard and methodology](https://www.vals.ai/benchmarks/science_index)

Vals released v1 on October 16, 2026. The index weights BioMysteryBench, MysteryMechanism and Terminal-Bench-Science equally and reports the mean accuracy. We mirror the published aggregates as a display-only composite, outside overall and category rankings.

Vals Science Index is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (23 models)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [GPT-6 Astra](/models/gpt-6-astra) | max reasoning | OpenAI | 65.09% |
| 2 | [GPT-6.1 Sol](/models/gpt-6-1-sol) | max reasoning | OpenAI | 59.63% |
| 3 | [Claude Sonnet 5.5](/models/claude-sonnet-5-5) | — | Anthropic | 59.12% |
| 4 | [Claude Opus 5.5](/models/claude-opus-5-5) | — | Anthropic | 58.65% |
| 5 | [Gemini 4 Argon](/models/gemini-4-argon) | high reasoning | Google | 55.36% |
| 6 | [Muse Wonder Full](/models/muse-wonder-full) | max reasoning | Meta | 48.54% |
| 7 | [Claude Opus 5](/models/claude-opus-5) | — | Anthropic | 47.93% |
| 8 | [GPT-6 Sol](/models/gpt-6-sol) | max reasoning | OpenAI | 45.00% |
| 9 | [Kimi K3.1 Pro](/models/kimi-3-1-pro) | max reasoning | Moonshot AI | 44.37% |
| 10 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | max reasoning | OpenAI | 41.48% |
| 11 | [Muse Wonder](/models/muse-wonder) | max reasoning | Meta | 39.12% |
| 12 | [Claude Haiku 5.5](/models/claude-haiku-5-5) | — | Anthropic | 38.50% |
| 13 | [Grok 4.6](/models/grok-4-6) | high reasoning | xAI | 38.09% |
| 14 | [Gemini 3.8 Flash](/models/gemini-3-8-flash) | high reasoning | Google | 35.76% |
| 15 | [Grok 4.7](/models/grok-4-7) | xhigh reasoning | xAI | 34.83% |
| 16 | [Ember-1](/models/ember-1) | — | Fireworks | 31.76% |
| 17 | [MiMo-V2.6-Flash](/models/mimo-v2-6-flash) | — | Xiaomi | 31.72% |
| 18 | [DeepSeek V4.1 Flash 1017](/models/deepseek-v4-1-flash) | high reasoning | DeepSeek | 29.65% |
| 19 | [Step 5 Preview](/models/step-5-preview) | — | StepFun | 29.58% |
| 20 | [GPT-6 Luna](/models/gpt-6-luna) | max reasoning | OpenAI | 28.38% |
| 21 | [Gemini 3.1 Pro Preview](https://www.vals.ai/models/google_gemini-3.1-pro-preview) | high reasoning | Google | 27.38% |
| 22 | [GPT-5.6 Luna](/models/gpt-5-6-luna) | max reasoning | OpenAI | 25.30% |
| 23 | [Beam 1.0](https://www.vals.ai/models/reflection_Beam-1.0) | — | Reflection | 22.03% |

## FAQ

### What does Vals Science Index measure?

Vals AI composite of agentic science work: investigating biology datasets, recovering hidden mechanisms through experiments, and running research workflows in a terminal.

### Which model leads the published Vals Science Index snapshot?

GPT-6 Astra currently leads the published Vals Science Index snapshot with a score of 65.09%.

### How many models are evaluated on Vals Science Index?

The October 25, 2026 contains 23 AI models.

### Does Vals Science Index affect BenchLM's overall score?

Not directly. Vals Science Index is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.