LLM Stats (SWE-bench Multilingual): leaderboard
LLM Stats compilation of vendor-reported SWE-bench Multilingual results: 300 GitHub issues across 9 programming languages (C, C++, Go, Java, JavaScript, TypeScript, PHP, Ruby, Rust), resolved when the model's patch passes the repository tests. Percent resolved, higher is better. Harnesses and attempts differ by vendor, unlike the swebench.com mini-SWE-agent board.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around February 2027. 51 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5.5 | 93.9 | #3 |
| 2 | Claude Sonnet 5.5 | 90.3 | #14 |
| 3 | Claude Mythos Preview | 87.3 | #5 |
| 4 | Claude Opus 4.8 | 84.4 | #39 |
| 5 | Claude Haiku 5.5 | 83.7 | #43 |
| 6 | Hy4 preview | 82.9 | #45 |
| 7 | Qwen3.8-Flash-Next | 81 | #76 |
| 8 | Laguna S 2.1 | 78.5 | #464 |
| 9 | Qwen 3.7 Max | 78.3 | #85 |
| 10 | Claude Sonnet 5 | 78.3 | #65 |
| 11 | Beam 1.0 | 78.3 | #102 |
| 12 | Claude Opus 4.6 | 77.83 | #67 |
| 13 | Kimi K2.6 | 76.7 | #113 |
| 14 | MiniMax-M2.7 | 76.5 | #270 |
| 15 | DeepSeek V4 Pro (Max) | 76.2 | #112 (DeepSeek V4 Pro) |
No result here: #6 GPT-6 Astra, #7 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-swe-bench-multilingual · How It Works · Data refreshed daily, snapshot 2026-11-05.