Claw-SWE-Bench (Hermes Agent): leaderboard
Claw-SWE-Bench (2026) compares general-purpose agent harnesses (claws) on SWE-bench-style issue resolution: 350 multilingual GitHub issues run under a fixed prompt, time budget, patch extraction and evaluator. This board is Pass@1 in percent with the Hermes Agent harness at high reasoning effort; higher is better.
Metric: Pass@1 (%) on the 350 GitHub issue-resolution instances of Claw-SWE-Bench (8 languages, 43 repositories, from SWE-bench-Multilingual and SWE-bench-Verified-Mini after future-commit cleanup); full adapter, identical prompt, 60-minute per-instance budget, one run, SWE-bench evaluator; hermes-agent harness; reasoning_effort high. Source: arxiv.org. Saturation forecast: Around February 2027. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GLM-5.1 | 71.1 | #179 |
| 2 | Qwen 3.6 Flash | 62.6 | #424 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=claw-swe-bench-hermes-agent · How It Works · Data refreshed daily, snapshot 2026-11-13.