SWE-bench Multimodal
GitHub issues in visual, user-facing JavaScript projects where the issue includes screenshots. Run by Vals AI.
| Category | Coding |
|---|---|
| Metric | Accuracy (%) |
| Source | Vals AI (evaluator) |
| License | None stated |
| Benchmark page | www.vals.ai |
| Status | Scored |
| Weight | 0.8 within Coding |
| Fit | Discrimination 0.60, noise 0.23 |
Results
| # | Model | Provider | Result | Date |
|---|---|---|---|---|
| 1 | Claude Opus 5.5 Max | Anthropic | 70.2% | 2026-11-09 |
| 2 | Claude Fable 5.2 High | Anthropic | 62.3% | 2026-11-09 |
| 3 | Claude Fable 5.1 Max | Anthropic | 61.3% | 2026-11-09 |
| 4 | Claude Sonnet 5.5 Max | Anthropic | 58.3% | 2026-11-09 |
| 5 | Gemini 4 Argon High | 56.0% | 2026-11-09 | |
| 6 | Muse Wonder Full Max | Meta | 47.3% | 2026-11-09 |
| 7 | DeepSeek V4.1 Flash Max | DeepSeek | 46.5% | 2026-11-09 |
| 8 | Muse Wonder Max | Meta | 45.6% | 2026-11-09 |
| 9 | Kimi K3 Max | Moonshot AI | 44.2% | 2026-11-09 |
| 10 | Claude Haiku 5.5 Max | Anthropic | 44.0% | 2026-11-09 |
| 11 | GPT-6.1 Sol Max | OpenAI | 43.5% | 2026-11-09 |
| 12 | GPT-6 Astra Max | OpenAI | 43.3% | 2026-11-09 |
| 13 | Inkling 2 Small | Thinking Machines | 43.1% | 2026-11-09 |
| 14 | Muse Spark 1.3 Max | Meta | 43.1% | 2026-11-09 |
| 15 | Grok 4.7 Extra High | xAI | 42.9% | 2026-11-09 |
| 16 | Qwen3.8 Max | Alibaba | 42.5% | 2026-11-09 |
| 17 | MiMo-V2.6-Pro | Xiaomi | 41.3% | 2026-11-09 |
| 18 | Step 5 Preview | StepFun | 40.8% | 2026-11-09 |
| 19 | Gemini 3.8 Flash High | 40.8% | 2026-11-09 | |
| 20 | GPT-6 Luna Max | OpenAI | 37.5% | 2026-11-09 |
| 21 | MiniMax-M3 | MiniMax | 33.3% | 2026-11-09 |
| 22 | Mistral Large 4 High | Mistral | 33.1% | 2026-11-09 |