Harbor Adapters - SWE-bench Verified (Terminus-2): leaderboard
SWE-bench Verified, the human-curated set of real Python GitHub issues, here 100 of its 500 tasks: the percentage of issues the agent's patch resolves, judged by the repository's unit tests, higher is better. Each model runs in Terminus-2, the Harbor framework's bash-only terminal agent with no native web or image tools, at its provider's default settings, so results can differ from the same model in its vendor's own agent; the score is the mean of three trials (Harbor Adapters large-scale evaluation; Shi, Lin et al., 2026).
Metric: Resolved rate (%; 100 tasks, mean of 3 trials). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 77.7 | #91 |
| 2 | Claude Opus 4.6 | 75 | #100 |
| 3 | Claude Sonnet 4.6 | 73.3 | #158 |
| 4 | GPT-5.4 (Medium) | 69 | #121 (GPT-5.4) |
| 5 | Gemini 3 Flash (Preview) | 68.7 | #141 |
| 6 | Claude Haiku 4.5 | 61.7 | #568 |
| 7 | GPT-5 Mini | 30.7 | #349 |
| 8 | GPT-5 Nano | 16 | #888 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=harbor-adapters-swe-bench-verified-terminus-2 · How It Works · Data refreshed daily, snapshot 2026-11-12.