HOLMES: leaderboard
Metric: Overall Accuracy (self-reported). Source: benchmarklist.com. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.6 Flash | 59.54 |
| 2 | MiniMax-M2.5 | 54.77 |
| 3 | Gemini 3.1 Flash | 54.31 |
| 4 | DeepSeek V4 Flash | 50.98 |
| 5 | DeepSeek R1 | 50.11 |
| 6 | DeepSeek V3.2 | 47.43 |
| 7 | MiniMax-M2.7 | 46.78 |
| 8 | GPT-5.4 Mini | 45.39 |
| 9 | DeepSeek V4 Pro | 44.49 |
Interactive version: theaggregate.ai/benchmark?slug=holmes · How It Works · Data refreshed daily, snapshot 2026-09-05.