SWEeper-Bench - Preservation: leaderboard
Metric: Preservation rate (%; 200 interactive software bug-discovery tasks; native agent harnesses; scoped prompt; unlimited time). Source: sweeperbench.com. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Grok 4.6 (xHigh) | 100 |
| 2 | GPT-6 Astra (xHigh) | 100 |
| 3 | Claude Fable 5.1 (xHigh) | 100 |
| 4 | Kimi K3 (High) | 99.5 |
| 5 | GLM-5.3 (Max) | 99 |
| 6 | GPT-5.6 Terra (xHigh) | 99 |
| 7 | GLM-5.3 Flash (Max) | 99 |
| 8 | GPT-5.6 Luna (xHigh) | 98.5 |
| 9 | Claude Opus 5 (xHigh) | 98.5 |
| 10 | GPT-5.6 Sol (xHigh) | 98 |
Interactive version: theaggregate.ai/benchmark?slug=sweeper-bench-preservation · How It Works · Data refreshed daily, snapshot 2026-10-09.