Active-SWE - Revealed: leaderboard
Metric: Potential Bugs Revealed (%). Source: hbinli.github.io. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.8 | 75 |
| 2 | GLM-5.2 | 69 |
| 3 | GPT-5.5 | 65.5 |
| 4 | Qwen 3.7 Max | 59.8 |
| 5 | Hy3 | 58.5 |
| 6 | Gemini 3.1 Pro (Preview) | 57 |
| 7 | Ring-2.6-1T | 56.3 |
| 8 | Claude Sonnet 4.6 | 53.8 |
| 9 | DeepSeek V4 Pro | 53.3 |
| 10 | Kimi K2.6 | 51.8 |
| 11 | Kimi K2.7 Code | 47.8 |
| 12 | Qwen 3.5 397B A17B | 42 |
| 13 | Qwen 3.5 122B A10B | 40 |
| 14 | Ling-3.0-flash | 38.75 |
| 15 | MiniMax-M2.7 | 38.5 |
Interactive version: theaggregate.ai/benchmark?slug=active-swe-revealed · How It Works · Data refreshed daily, snapshot 2026-09-19.