SWE-bench Live Windows (50 tasks, Win-agent): leaderboard
SWE-bench-Live's Windows split (Microsoft, 2026 update): recent GitHub issues whose fixes must be made and tested inside Windows containers through PowerShell, graded by the repository's tests. This board is the 50-task snapshot under Win-agent, the maintainers' minimal Windows-compatible agent with SWE-agent's tool calls, scored as the percentage of tasks resolved; higher is better. The split grows, so each snapshot size is its own board.
Metric: Resolved rate (%). Source: swe-bench-live.github.io. Saturation forecast: Around July 2028. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.5 | 30 | #244 |
| 2 | GPT-5.2 (Medium) | 20 | #172 (GPT-5.2) |
| 3 | DeepSeek V3.1 Terminus | 20 | #442 |
| 4 | Gemini 3 Flash | 16 | #159 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=swe-bench-live-windows-50-tasks-win-agent · How It Works · Data refreshed daily, snapshot 2026-11-12.