LiveBench JavaScript: leaderboard
The JavaScript task in LiveBench's agentic coding category: the model works in a terminal environment to resolve a real issue from a JavaScript GitHub repository. Runs use the Mini-SWE-Agent scaffold with a 250 step limit, and a patch counts only when the repository tests pass under the Multi-SWE-Bench harness. The board shows the task score in percent, higher is better. Source: github.com/LiveBench/LiveBench.
Metric: Score. Source: livebench.ai. Saturation forecast: Around November 2027. 71 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4.1 Flash (Max) | 81.82 | #36 (DeepSeek V4.1 Flash) |
| 2 | Claude Opus 5.5 (Max) | 81.82 | #3 (Claude Opus 5.5) |
| 3 | Qwen 3.8 Max | 77.27 | #35 |
| 4 | Claude Opus 5 (Max) | 77.27 | #12 (Claude Opus 5) |
| 5 | Muse Spark 1.3 (xHigh) | 77.27 | #24 (Muse Spark 1.3) |
| 6 | Muse Spark 1.1 (xHigh) | 77.27 | #52 (Muse Spark 1.1) |
| 7 | Claude Sonnet 5.5 (Max) | 77.27 | #11 (Claude Sonnet 5.5) |
| 8 | GLM-5.3 | 72.73 | #39 |
| 9 | Grok 4.5 | 72.73 | #48 |
| 10 | Gemini 3.8 Flash (High) | 72.73 | #21 (Gemini 3.8 Flash) |
| 11 | Grok 4.6 | 72.73 | #26 |
| 12 | Muse Spark 1.2 (xHigh) | 72.73 | #62 (Muse Spark 1.2) |
| 13 | Claude Opus 5.5 (xHigh) | 72.73 | #3 (Claude Opus 5.5) |
| 14 | Kimi K3 | 68.18 | #30 |
| 15 | Claude Opus 4.8 (Max) | 68.18 | #37 (Claude Opus 4.8) |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=livebench-javascript · How It Works · Data refreshed daily, snapshot 2026-10-20.