Open Source Isaac 0.5: 36B Embodied Model Trained on Web Videos
lukas_m_ziegler · x · 2026-08-27
Perceptron Inc (ex-Meta Chameleon team) released Isaac 0.5, an open-source embodied foundation model that turns internet video into robot training data. The same weights handle video QA, object pointing, tracking, task reporting, and action generation.
Key specs:
- Scale: 36B-parameter sparse model, trained on 35+ robot systems and 3T multimodal tokens.
- Performance: Scores 62.6 on ScreenSpot-Pro grounding vs 54.8 for the strongest Qwen3-VL.
- Efficiency: Achieves this at 8.5× lower inference cost (26.9 TFLOP vs 228.4).
- Tech: Uses "Null Experts" where each token selects how many experts to activate, allocating compute dynamically.
Related event: Perceptron releases Isaac 0.5, a 36B open-weight embodied foundation model(10 posts)→
More from Embodied
- SEAR: first large-scale controlled study of self-evolving robot agents, 7 LLMs, 330 tasks — xiye_nlp · 2026-10-16
- Podcast: subsea robotics startup Ulysses on operating 1,500m underwater and ocean carbon — blakemenezes · 2026-10-16
- Sony unveils $5,900 gold-plated Walkman with AI music enhancement — Polymarket · 2026-10-16
- ROBIB Meetup #3 returns to Ghent on Nov 19 with Physical AI talks, demos and panel — MarwaEldiwiny · 2026-10-16
- VLX-Go: Multi-Stage Vision-Language Training Boosts Robot Person-Following Among Distractors — omlab · 2026-10-16
- Ledger Atom Debuts as Agent-Approval Security Device; Critic Says a $100 iPhone 8 Beats It — adamamcbride · 2026-10-16