AppWorld Normal: leaderboard
Interactive coding agent benchmark simulating 9 everyday apps (email, calendar, Spotify, etc.) with 457 APIs and 750+ test scenarios. Normal difficulty split.
Metric: Task Goal Completion (%). Source: appworld.dev. Source: appworld.dev. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 91.1 | #63 |
| 2 | Claude Opus 4.7 | 88.7 | #49 |
| 3 | Qwen 3 14B | 86.9 | #537 |
| 4 | GPT-5.6 Luna | 85.1 | #68 |
| 5 | GPT-4.1 | 73.2 | #248 |
| 6 | Qwen 2.5 32B | 72.6 | #546 |
| 7 | GPT-4o | 68.5 | #343 |
| 8 | GPT-4.1 Mini | 61.9 | #355 |
| 9 | GPT-4 Turbo | 32.7 | #404 |
| 10 | Llama 3 70B | 24.4 | #631 |
Interactive version: theaggregate.ai/benchmark?slug=appworld-normal · How It Works · Data refreshed daily, snapshot 2026-10-20.