Next.js AI Agent Evaluations
Agents · 2025-10-21
Next.js AI Agent Evaluations measures framework-specific generation, migration, caching, routing, rendering, and best-practice tasks under paired runs with and without bundled documentation supplied through AGENTS.md. The publisher changes the task roster over time: the October 16, 2026 snapshot contains 31 tasks, so this evolving leaderboard has no fixed canonical item count.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5.5 | 97.0 |
| GPT-6 Sol | 97.0 |
| Gemini 4 Argon | 97.0 |
| Claude Fable 5.1 | 97.0 |
| Gemini 3.8 Flash | 97.0 |
| GPT-6 Astra | 97.0 |
| Kimi K3 | 97.0 |
| GLM-5.2 | 97.0 |
| Sonnet 5 | 97.0 |
| Claude Opus 5 | 97.0 |
| Grok 4.7 | 94.0 |
| Composer 2.5 | 94.0 |
| MiniMax M3 | 90.0 |
| Kimi K2.7 Code | 90.0 |
| GPT-6 Luna | 87.0 |