Agent-ValueBench: leaderboard
Agent value-alignment benchmark with executable environments and value-conflict tasks, testing whether autonomous agents express stable values across domains, harnesses, and trajectories.
Metric: Authority (self-reported). Source: huggingface.co. Source: benchmarklist.com. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 30B A3B | 7.8 | #493 |
| 2 | Grok 4.20 | 7.8 | #175 |
| 3 | Llama 3.3 70B Instruct | 7.5 | #534 |
| 4 | Gemini 3.1 Pro (Preview) | 6.7 | #58 |
| 5 | MiniMax-M2.7 | 5.6 | #258 |
| 6 | GPT-5.4 Mini | 5.5 | #212 |
| 7 | Claude Sonnet 4.6 | 5 | #90 |
| 8 | Claude Haiku 4.5 | 4.7 | #280 |
| 9 | Qwen 3.5 397B A17B | 4.7 | #146 |
| 10 | GPT-5.4 | 4.6 | #79 |
| 11 | GLM-5.1 | 4.4 | #97 |
| 12 | Kimi K2.5 | 3.7 | #144 |
| 13 | DeepSeek V3.2 | 3.4 | #205 |
| 14 | Gemini 3 Flash (Preview) | 2.1 | #82 |
Interactive version: theaggregate.ai/benchmark?slug=agent-valuebench · How It Works · Data refreshed daily, snapshot 2026-10-20.