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: benchmarklist.com. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3 30B A3B | 7.8 |
| 2 | Grok 4.20 | 7.8 |
| 3 | Llama 3.3 70B Instruct | 7.5 |
| 4 | Gemini 3.1 Pro (Preview) | 6.7 |
| 5 | MiniMax-M2.7 | 5.6 |
| 6 | GPT-5.4 Mini | 5.5 |
| 7 | Claude Sonnet 4.6 | 5 |
| 8 | Claude Haiku 4.5 | 4.7 |
| 9 | Qwen 3.5 397B A17B | 4.7 |
| 10 | GPT-5.4 | 4.6 |
| 11 | GLM-5.1 | 4.4 |
| 12 | DeepSeek V3.2 | 3.4 |
| 13 | Gemini 3 Flash (Preview) | 2.1 |
Interactive version: theaggregate.ai/benchmark?slug=agent-valuebench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.