GenXR-Bench - App Generation: leaderboard
The 67 GenXR-Bench tasks that ask the agent to build a complete XR application from a natural-language specification; the score is the share whose held-out interaction tests all pass in one attempt.
Metric: Pass@1 on the 67 application-generation tasks (%). Source: arxiv.org. Saturation forecast: Around March 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 (Medium) | 47.8 | #22 (Claude Opus 5) |
| 2 | Gemini 3.8 Flash (High) | 34.3 | #36 (Gemini 3.8 Flash) |
| 3 | GPT-5.6 Sol (xHigh) | 34.3 | #30 (GPT-5.6 Sol) |
| 4 | Gemini 3.7 Flash (High) | 29.9 | #32 (Gemini 3.7 Flash) |
| 5 | Gemini 3.5 Flash (High) | 17.9 | #101 (Gemini 3.5 Flash) |
| 6 | Gemini 3.6 Flash (High) | 17.9 | #76 (Gemini 3.6 Flash) |
| 7 | GPT-5.6 Luna (xHigh) | 17.9 | #121 (GPT-5.6 Luna) |
| 8 | Gemini 3.1 Pro (Preview) (High) | 9 | #94 (Gemini 3.1 Pro (Preview)) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=genxr-bench-app-generation · How It Works · Data refreshed daily, snapshot 2026-11-13.