React Native Evals: leaderboard
Callstack's open React Native coding suite; the board covers 91 evals in six groups (navigation, animation, async state, lists, React Native APIs, Skia) out of 134 in the repository. Each eval hands the model scaffold files and a task prompt and takes back modified files, which an LLM judge checks requirement by requirement. The score is the share of requirements met across ten runs, and higher is better.
Metric: Overall Score (%). Source: rn-evals.vercel.app. Saturation forecast: Around January 2027. 32 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 | 89.05 | #9 |
| 2 | Claude Opus 5.5 | 88.57 | #3 |
| 3 | Claude Opus 5 | 88.12 | #12 |
| 4 | Claude Fable 5 | 86.96 | #16 |
| 5 | GPT-6 Astra | 86.37 | #6 |
| 6 | GPT-5.6 Sol | 86.34 | #19 |
| 7 | GPT-6.1 Sol | 86.34 | #7 |
| 8 | Claude Opus 4.8 | 84.69 | #37 |
| 9 | GPT-5.5 | 84.65 | #31 |
| 10 | Grok 4.5 | 84.58 | #48 |
| 11 | Grok 4.6 | 84.07 | #26 |
| 12 | GPT-5.6 Terra | 83.52 | #41 |
| 13 | Claude Sonnet 5 | 83.41 | #57 |
| 14 | GPT-6 Sol | 83.25 | #14 |
| 15 | GPT-5.6 Luna | 82.86 | #68 |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=react-native-evals · How It Works · Data refreshed daily, snapshot 2026-10-20.