Truthful AI Advisors (Crawford-Sobel cheap-tal: leaderboard
Large language models are increasingly deployed as advisors whose objective is not aligned with the user's: recommenders optimize for engagement, sales assistants for purchases, negotiation agents for concessions. Whether such advisors stay truthful when honesty conflicts with their own payoff is a core alignment-evaluation question. We turn the canonical Crawford-Sobel cheap-talk model into a pre
Metric: Mean Normalized MI (positive bias) (self-reported). Source: benchmarklist.com. Saturation forecast: Around February 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 96.9 | #142 |
| 2 | Llama 3.3 70B | 92.6 | #579 |
| 3 | GPT-4o | 88.1 | #343 |
| 4 | Gemini 2.5 Flash Lite | 77.6 | #422 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=truthful-ai-advisors-crawford-sobel-cheap-tal · How It Works · Data refreshed daily, snapshot 2026-10-20.