MultiView-Bench - 3D DoF 1: leaderboard
Metric: Accuracy (%; six views, objects share the central object axis; fraction of 100 procedurally generated Blender scenes whose relative position of a target object is answered exactly as (+/-X or 0, +/-Y or 0, +/-Z or 0) in a visible world coordinate frame; mean over three runs where printed (Claude 3.7 Sonnet one run); x100). Source: arxiv.org. Saturation forecast: Around July 2027. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.6 Luna (Medium) | 63.7 |
| 2 | Gemini 3.1 Pro (Preview) (Medium) | 56.3 |
| 3 | O3 | 52.3 |
| 4 | GPT-5.6 Sol (Medium) | 52.3 |
| 5 | GPT-5.6 Terra (Medium) | 42.7 |
| 6 | Claude 3.7 Sonnet | 42 |
| 7 | GPT-5 | 39.7 |
| 8 | Gemini 3.5 Flash (Medium) | 35 |
| 9 | GPT-4o | 27.3 |
| 10 | Gemini 2.5 Pro | 26 |
| 11 | Gemini 2.5 Flash | 21.7 |
| 12 | Claude Haiku 4.5 (Thinking) | 17 |
| 13 | Claude Sonnet 4 | 10.3 |
Interactive version: theaggregate.ai/benchmark?slug=multiview-bench-3d-dof-1 · How It Works · Data refreshed daily, snapshot 2026-09-29.