WEAVE-Bench (Cline) - Resolved: leaderboard
WEAVE-Bench dependency-structured multi-PR repository tasks; Cline agent scaffold, shell and file-edit tools, at most 200 interaction turns, 2-hour task and 30-minute test-suite timeouts, provider-default context window and temperature. Resolved: all fail-to-pass tests pass with no pass-to-pass regressions. Model labels as printed; checkpoint dates and reasoning settings are not disclosed. Metric: Resolved rate (%). Higher is better. Source: https://arxiv.org/html/2610.22039v1
Metric: Resolved rate (%). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 48.7 | #50 |
| 2 | GLM-5.2 | 43 | #71 |
| 3 | Kimi K2.6 | 37.3 | #105 |
| 4 | Claude Opus 4.6 | 36.3 | #62 |
| 5 | DeepSeek V4 Flash | 34.3 | #134 |
| 6 | GLM-5.1 | 21 | #94 |
| 7 | Kimi K2.5 | 18.7 | #143 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=weave-bench-cline-resolved · How It Works · Data refreshed daily, snapshot 2026-10-18.