WebApp1K Duo: leaderboard
React web app code generation benchmark (hard): dual-task combinations requiring models to implement two features simultaneously, testing compositional code generation.
Metric: Pass@1 (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 49 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 79.38 | #98 |
| 2 | Claude Sonnet 4.5 | 78.4 | #142 |
| 3 | Claude Opus 4.1 (20250805) | 76.9 | #132 |
| 4 | GPT-5.1 | 76.5 | #135 |
| 5 | Claude Opus 4 (20250514) | 76.3 | #154 |
| 6 | Claude Opus 4.5 (20251101) | 76.3 | #74 |
| 7 | GPT-OSS-120B | 75 | #341 |
| 8 | Claude 3.5 Sonnet (20241022) | 75 | #291 |
| 9 | Gemini 3 Pro (Preview) | 75 | #66 |
| 10 | GPT-5 Codex | 75 | #127 |
| 11 | GPT-5.1 Codex | 74.3 | #159 |
| 12 | O4 Mini | 72.6 | #180 |
| 13 | Grok 4.1 Fast (Reasoning) | 71.6 | #217 (Grok 4.1 Fast) |
| 14 | Kimi K2 (Thinking) | 70.7 | #243 (Kimi K2) |
| 15 | Gemini 2.5 Pro (Preview 06-05) | 70.4 | #106 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=webapp1k-duo · How It Works · Data refreshed daily, snapshot 2026-10-20.