PAIR-Bench - Hint Efficiency: leaderboard
Hint efficiency on PAIR-Bench (2026), from 0 to 6: seven minus the hint depth at which each attempted failure scenario is closed, 0 when it is never closed, averaged over scenarios. Higher means weaker, less explicit hints sufficed.
Metric: Hint efficiency score (out of 6): 7 minus the hint level, from 1 for coarse symptoms to 6 for implementation guidance, at which each attempted failure scenario is closed, 0 when it is never closed, averaged over scenarios and initially failing instances; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V3.2 | 4.79 | #406 |
| 2 | GPT-4o Mini | 3.8 | #1623 |
| 3 | Gemini 2.5 Flash Lite | 3.76 | #958 |
| 4 | Qwen 3 Coder 30B A3B Instruct | 3.61 | #972 |
| 5 | Ministral 3 14B | 3.29 | #1845 |
| 6 | Llama 3.3 70B Instruct | 3.21 | #1357 |
| 7 | Mistral Small 3.2 | 3.02 | #1591 |
| 8 | Gemma 3 27B | 2.76 | #1631 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=pair-bench-hint-efficiency · How It Works · Data refreshed daily, snapshot 2026-11-13.