ReactBench: leaderboard
ReactBench evaluates large language model agents on realistic React UI implementation tasks. The inputs are text. The pinned metric is Pass@1 (%), measuring first-attempt success, with the source also reporting 95% confidence intervals. Source: BenchmarkList’s attributed result table.
Metric: Pass@1 (%). Source: benchmarklist.com. Saturation forecast: Around September 2028. 72 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol (Max) | 46.7 | #22 (GPT-5.6 Sol) |
| 2 | Claude Fable 5.1 (Max) | 45.6 | #10 (Claude Fable 5.1) |
| 3 | Claude Fable 5.1 (xHigh) | 44.1 | #10 (Claude Fable 5.1) |
| 4 | Claude Fable 5 (Max) | 43.6 | #20 (Claude Fable 5) |
| 5 | Claude Opus 5 (Max) | 42.1 | #16 (Claude Opus 5) |
| 6 | GPT-5.6 Sol (xHigh) | 42.1 | #22 (GPT-5.6 Sol) |
| 7 | Claude Opus 5 (High) | 42.1 | #16 (Claude Opus 5) |
| 8 | Claude Fable 5 (xHigh) | 42.1 | #20 (Claude Fable 5) |
| 9 | GPT-5.6 Terra (Max) | 40.5 | #40 (GPT-5.6 Terra) |
| 10 | Claude Fable 5.1 (High) | 40.5 | #10 (Claude Fable 5.1) |
| 11 | Claude Fable 5 (High) | 39 | #20 (Claude Fable 5) |
| 12 | GPT-5.6 Sol (High) | 37.4 | #22 (GPT-5.6 Sol) |
| 13 | GPT-5.6 Terra (xHigh) | 37.4 | #40 (GPT-5.6 Terra) |
| 14 | GPT-5.6 Terra (High) | 36.9 | #40 (GPT-5.6 Terra) |
| 15 | Claude Opus 5 (xHigh) | 36.4 | #16 (Claude Opus 5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=reactbench · How It Works · Data refreshed daily, snapshot 2026-11-05.