EmbodimentSemantic - Agentview: leaderboard

Metric: Mean per-task triplet F1 (%; directed object-relation-object triplets predicted from single LIBERO-Spatial agentview (fixed third-person camera) frames against simulator-derived ground-truth scene graphs over a fixed vocabulary of 7 objects and 8 directed relations; exact triplet F1, invariant to swapping the two identical bowls, averaged over the 10 manipulation tasks; one fixed prompt, valid triplets only; higher is better). Source: arxiv.org. Saturation forecast: Around January 2027. 10 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)56.74
2InternVL3-78B35.19
3Qwen 3 VL 8B Instruct28.37
4gemma-4-E4B-it22.09
5Molmo2-8B18.88
6InternVL3.5-8B17.69
7Nemotron Nano 12B v2 VL15.99
8Qwen 2.5 VL 7B Instruct14.05

Interactive version: theaggregate.ai/benchmark?slug=embodimentsemantic-agentview · How It Works · Data refreshed daily, snapshot 2026-09-29.