SWE-bench Live Java (62 tasks, SWE-agent): leaderboard
SWE-bench-Live's Java split (Microsoft, 2026 multi-language update): recent GitHub issues from Java repositories, built into runnable environments by the RepoLaunch pipeline and graded by the repository's tests. This board is the 62-task snapshot under the SWE-agent harness, 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 August 2028. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V3.1 Terminus | 27.42 | #442 |
| 2 | Gemini 3 Flash | 25.81 | #159 |
| 3 | GPT-5.2 (Medium) | 22.58 | #172 (GPT-5.2) |
| 4 | Claude Sonnet 4.5 | 16.13 | #244 |
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-java-62-tasks-swe-agent · How It Works · Data refreshed daily, snapshot 2026-11-12.