GroundBench - 64 Vertices: leaderboard
Metric: Filled-region IoU (0-100), IoU x 100 (N=64 vertex budget; 1,500 RefCOCO, RefCOCO+ and RefCOCOg image, expression and referent triples (500 each) re-targeted to exact-N polygons; the model returns exactly N integer vertices and the filled polygon is scored by IoU against the canonical target; fixed denominator (an unparseable answer scores 0); first parseable output over one call plus up to two stateless retries). Source: arxiv.org. Saturation forecast: Around 2028. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.5 Flash (High) | 37.7 |
| 2 | Seed 2.0 Pro | 37.6 |
| 3 | Kimi K2.7 Code | 24.4 |
Interactive version: theaggregate.ai/benchmark?slug=groundbench-64-vertices · How It Works · Data refreshed daily, snapshot 2026-09-26.