LLMConfBench: leaderboard
Metric: Conformer Ranking Accuracy (mean reliable-pair Kendall tau). Source: arxiv.org. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-6 | 0.67 |
| 2 | Claude Opus 5 | 0.61 |
| 3 | Gemini 3.6 Flash | 0.43 |
| 4 | GPT-5.6 Sol | 0.4 |
| 5 | Kimi K3 | 0.38 |
| 6 | Claude Sonnet 5 | 0.37 |
| 7 | DeepSeek V4 Flash (0731) | 0.33 |
| 8 | Gemini 3 Flash (Preview) | 0.32 |
| 9 | Qwen 3.7 Max | 0.32 |
| 10 | Qwen 3.8 27B | 0.31 |
| 11 | GPT-5 | 0.3 |
| 12 | Gemma 4 31B | 0.27 |
| 13 | Kimi K2 (Thinking) | 0.27 |
| 14 | Qwen 3.5 27B | 0.19 |
| 15 | Qwen 3.5 Plus (2026-04-20) | 0.18 |
Interactive version: theaggregate.ai/benchmark?slug=llmconfbench · How It Works · Data refreshed daily, snapshot 2026-09-20.