CLBench - Empirical Discovery & Simulation: leaderboard
Metric: Solving Rate (%). Source: www.clbench.com. 36 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 (xHigh) | 22.8 |
| 2 | Hy3-preview | 19.6 |
| 3 | Kimi K2.5 | 18.1 |
| 4 | GPT-5.1 (High) | 18.1 |
| 5 | Seed 2.0 Pro (High) | 17.2 |
| 6 | Qwen 3.5 Plus (Thinking) | 16.7 |
| 7 | Claude Opus 4.5 (Thinking) | 15.1 |
| 8 | Gemini 3.1 Pro (Preview) (High) | 14.5 |
| 9 | Seed 2.0 Pro (Medium) | 13.9 |
| 10 | O3 (High) | 13.7 |
| 11 | GPT-5.1 | 13.6 |
| 12 | Qwen 3.6 Plus | 13.5 |
| 13 | Kimi K2 (Thinking) | 12.6 |
| 14 | Grok 4.20 (Reasoning) | 12.3 |
| 15 | GPT-5.2 | 12.1 |
Interactive version: theaggregate.ai/benchmark?slug=clbench-empirical-discovery-simulation · How It Works · Data refreshed daily, snapshot 2026-09-19.