# Vals RSI Index v1.1 (Vals RSI Index)

> Tests autonomous AI research on compression, language-model training, judge-harness engineering, and post-training under fixed compute and time budgets.

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

- Category: [Agentic](/agentic)
- Last updated: October 15, 2026

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

## About Vals RSI Index

- Year: 2026
- Tasks: 4 autonomous research tasks
- Format: Mean reference-anchored log score, expressed as a percentage
- Difficulty: AI research under fixed compute and time budgets
- Paper: [Vals RSI Index leaderboard and methodology](https://www.vals.ai/benchmarks/rsi_index)

The published composite averages four reference-anchored log scores. On the percentage scale, 0% marks a task baseline, 50% matches a selected published, human, or frontier-model reference, and 100% marks the theoretical best. These are index points rather than task success rates. Models use provider-specific agent harnesses at their highest supported reasoning effort. We keep every published row display only and outside overall and category rankings.

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

## Leaderboard (25 models)

| Rank | Model | Configuration | Creator | Score |
|------|-------|---------------|---------|-------|
| 1 | [Claude Opus 5.5](/models/claude-opus-5-5) | max reasoning | Anthropic | 37.31% |
| 2 | [Claude Fable 5.1](/models/claude-fable-5-1) | max reasoning | Anthropic | 36.09% |
| 3 | [Claude Opus 5](/models/claude-opus-5) | max reasoning | Anthropic | 33.02% |
| 4 | [Gemini 4 Argon](/models/gemini-4-argon) | high reasoning | Google | 30.55% |
| 5 | [GPT-6 Sol](/models/gpt-6-sol) | max reasoning | OpenAI | 28.11% |
| 6 | [GPT-6 Astra](/models/gpt-6-astra) | max reasoning | OpenAI | 27.07% |
| 7 | [GPT-6.1 Sol](/models/gpt-6-1-sol) | max reasoning | OpenAI | 26.62% |
| 8 | [Grok 4.6](/models/grok-4-6) | xhigh reasoning | xAI | 25.46% |
| 9 | [Muse Wonder Full](/models/muse-wonder-full) | max reasoning | Meta | 24.43% |
| 10 | [Claude Fable 5](/models/claude-fable) | max reasoning | Anthropic | 24.32% |
| 11 | [Claude Opus 4.8](/models/claude-opus-4-8) | max reasoning | Anthropic | 23.98% |
| 12 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | max reasoning | OpenAI | 23.88% |
| 13 | [GLM-5.3](/models/glm-5-3) | max reasoning | Z.AI | 22.91% |
| 14 | [Qwen3.8 Max](/models/qwen3-8-max) | max reasoning | Alibaba | 21.76% |
| 15 | [Kimi K3](/models/kimi-k3) | max reasoning | Moonshot AI | 20.88% |
| 16 | [Grok 4.7](/models/grok-4-7) | xhigh reasoning | xAI | 20.20% |
| 17 | [Gemini 3.8 Flash](/models/gemini-3-8-flash) | high reasoning | Google | 19.93% |
| 18 | [Muse Spark 1.3 Max](https://www.vals.ai/models/meta_muse_spark_1_3_max) | max reasoning | Meta | 19.64% |
| 19 | [Claude Opus 4.7](/models/claude-opus-4-7) | max reasoning | Anthropic | 19.08% |
| 20 | [Gemini 3.7 Flash](/models/gemini-3-7-flash) | high reasoning | Google | 18.27% |
| 21 | [GPT-5.5](/models/gpt-5-5) | xhigh reasoning | OpenAI | 16.48% |
| 22 | [GPT-5.2](/models/gpt-5-2) | xhigh reasoning | OpenAI | 15.80% |
| 23 | [GPT-5.4](/models/gpt-5-4) | xhigh reasoning | OpenAI | 15.48% |
| 24 | [Claude Opus 4.5](/models/claude-opus-4-5) | high reasoning | Anthropic | 14.82% |
| 25 | [Kimi K2.5 Thinking](https://www.vals.ai/models/kimi_kimi-k2.5-thinking) | max reasoning | Moonshot AI | 6.85% |

## FAQ

### What does Vals RSI Index measure?

Tests autonomous AI research on compression, language-model training, judge-harness engineering, and post-training under fixed compute and time budgets.

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

Claude Opus 5.5 currently leads the published Vals RSI Index snapshot with a score of 37.31%.

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

The October 15, 2026 contains 25 AI models.

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

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