Vals AI Public Benefits Bench: leaderboard
Metric: Accuracy (%). Source: www.vals.ai. 33 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 5 (Max) | 76.93 |
| 2 | Claude Fable 5.1 (Max) | 74.9 |
| 3 | Claude Fable 5 (Max) | 70.43 |
| 4 | GLM-5.3 (Max) | 68.54 |
| 5 | Muse Spark 1.2 (xHigh) | 68.47 |
| 6 | Kimi K3 | 68.27 |
| 7 | Claude Opus 4.8 (Max) | 68.13 |
| 8 | Qwen 3.8 Max | 67.12 |
| 9 | Grok 4.6 (High) | 66.85 |
| 10 | GPT-5.6 Sol (Max) | 66.51 |
| 11 | Claude Sonnet 5 (Max) | 66.03 |
| 12 | Gemini 3.8 Flash (High) | 65.29 |
| 13 | MiniMax-M3 | 64.14 |
| 14 | DeepSeek V4 Pro (Max) | 62.92 |
| 15 | Claude Sonnet 4.6 (Max) | 62.45 |
Interactive version: theaggregate.ai/benchmark?slug=vals-ai-public-benefits-bench · How It Works · Data refreshed daily, snapshot 2026-09-05.