Mind2Act World (Agents) - Overall Score: leaderboard

Diagnostic embodied manipulation benchmark (arXiv:2611.26444v1, Table 4) evaluating zero-shot multimodal API agents (Medium reasoning effort, one fixed motion-tool harness, 900 episodes each) across 30 tasks in Mind, Act, and Mind2Act views. Scored by weighted overall task score (0.25*Mind + 0.25*Act + 0.50*Mind2Act, 0-100).

Metric: Weighted overall task score (0-100). Source: arxiv.org. Saturation forecast: Around 2029. 3 models tracked. Updated 21 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-6 AstraMedium30.2#10
2DeepSeek V4.1 FlashMedium6.1#67
3Gemini 4 ArgonMedium4.4#19

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=mind2act-world-agents-overall-score · How It Works · Data refreshed daily, snapshot 2026-11-21.