ReactBench v1
Code · 2026-07-14
ReactBench v1 evaluates coding agents on 51 realistic React tasks from 47 open-source repositories. It combines 27 Write React and 24 Fix React tasks, grading each rollout with hidden behavioral tests and a pinned React Doctor gate; published scores average pass@1 over five trials per task.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.6 Terra | 53.3 |
| GPT-5.6 Sol | 52.5 |
| Claude Opus 5 | 49.0 |
| Claude Fable 5 | 47.5 |
| GPT-5.6 Luna | 43.9 |
| Grok 4.5 | 40.4 |
| Opus 4.8 | 36.5 |
| Kimi K3 | 32.9 |
| GLM-5.2 | 32.5 |
| Sonnet 5 | 30.6 |
| Muse Spark 1.1 | 29.0 |
| Gemini 3.5 Flash | 26.3 |
| Gemini 3.1 Pro Preview | 23.5 |
| Kimi K2.7 Code | 23.5 |
| Composer 2.5 | 21.2 |
| Inkling | 11.0 |