Benchmarks · composite
Vals RSI Index v1.1: Harness Engineering: Judge
Can a model do the research that builds the next model? Autonomous LLM R&D scored against frozen published, human, and frontier-model references. Published Harness Engineering: Judge score in percentage points.
21 models · 21 results · updated 2026-10-15 · vals.ai ↗
Leaderboard
GPT-6 Sol max
More from this benchmark
6 published measuresScore vs any published measure6 measures: sub-scores, splits, counts
Breakdown
darker is better within each column · click a column to sort| Model | |||||
|---|---|---|---|---|---|
| GPT-6 Sol max | 819.2M | 552k | 310k | 817.3M | 2.0M |
| GPT-6.1 Sol max | 93.3M | 906k | 747k | 87.7M | 5.2M |
| Gemini 4 Argon max | 170.0M | 1.5M | 776k | 146.1M | 0 |
| Grok 4.7 max | 163.3M | 825k | 500k | 149.4M | 0 |
All results
21 rows| # | Model | Effort | Score |
|---|---|---|---|
| 1 | GPT-6 Sol | max | 22.8% |
| 2 | Fable 5.1 | max | 22.6% |
| 3 | Grok 4.6 | max | 21.8% |
| 4 | GPT-6.1 Sol | max | 20.1% |
| 5 | GPT-5.6 Sol | max | 18.5% |
| 6 | Fable 5 | max | 18.1% |
| 7 | GLM 5.3 | max | 17.4% |
| 8 | Opus 4.8 | max | 16.8% |
| 9 | Opus 5 | max | 14.8% |
| 10 | Opus 5.5 | max | 14.5% |
| 11 | Gemini 3.8 Flash | max | 14.5% |
| 12 | Muse Spark 1.3 | max | 14.5% |
| 13 | Gemini 4 Argon | max | 13.9% |
| 14 | Qwen 3.8 Max | max | 13.5% |
| 15 | Grok 4.7 | max | 11.3% |
| 16 | Kimi K3 | max | 9.9% |
| 17 | GPT-6 Astra | max | 9.5% |
| 18 | GPT-5.4 | xhigh | 8.7% |
| 19 | Gemini 3.7 Flash | max | 8.4% |
| 20 | Opus 4.7 | max | 5.6% |
| 21 | GPT-5.5 | xhigh | 5.6% |