SWE-bench Verified (Vals 15 min–1 hour split)
Code · 2026-06-17
Vals difficulty split containing SWE-bench Verified tasks estimated to take 15 minutes to 1 hour for humans. It preserves the duration split rather than aggregating all rows under the parent benchmark.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 96.9 |
| GPT-5.6 Sol | 95.0 |
| Claude Fable 5 | 94.6 |
| Kimi K3 | 94.6 |
| GPT-5.6 Luna | 92.0 |
| Opus 4.8 | 88.1 |
| Grok 4.5 | 83.5 |
| Muse Spark 1.1 | 82.4 |
| GLM-5.2 | 81.6 |
| GPT-5.5 | 81.2 |
| Gemini 3.6 Flash | 79.3 |
| Opus 4.7 | 78.9 |
| Composer 2.5 | 78.9 |
| Gemini 3.1 Pro Preview | 77.8 |
| Gemini 3.5 Flash | 77.8 |
| Sonnet 5 | 77.4 |
| GPT-5.4 | 76.2 |
| DeepSeek-V4-Pro | 75.9 |
| Opus 4.6 | 75.5 |
| GPT-5.6 Terra | 75.5 |
| Kimi K2.7 Code | 75.5 |
| Sonnet 4.6 | 75.1 |
| Inkling | 75.1 |
| Opus 4.5 | 74.7 |
| MiMo-V2.5-Pro | 74.7 |
| Kimi K2.6 | 74.3 |
| Gemini 3 Pro Preview | 73.6 |
| Gemini 3.5 Flash-Lite | 73.2 |
| GPT-5.3-Codex | 73.2 |
| GLM-5.1 | 72.8 |
| MiniMax M2.1 | 72.8 |
| MiniMax M3 | 72.8 |
| Muse Spark | 72.8 |
| GPT-5.2 | 72.4 |
| Gemini 3 Flash Preview | 72.0 |
| MiniMax M2.7 | 72.0 |
| GPT-5.4 Mini | 70.9 |
| MiniMax M2.5 | 70.9 |
| Qwen3.6 Plus (2026-04-02) | 70.9 |
| MiMo-V2.5 | 70.5 |