Perceptron releases Mk1.5, a 35B embodied-agent model with up to 4.7x faster task completion
rohanpaul_ai · x · 2026-09-26
Embodied-AI startup Perceptron released Mk1.5, a 35B embodied-agent model that can fly drones, control quadrupeds, power smart glasses, track objects, search the web, reason visually, and dispatch its own sub-agents — all without platform-specific retraining, with up to 4.7x faster request completion.
Key points:
- First-person video is a distinct problem: on robots or smart glasses, hands constantly enter and leave the frame, occlude objects, and the camera itself moves. Mk1.5 was trained specifically on egocentric video.
- Tool use delivers big gains: on MMSearch, F1 jumps from 18.2 to 54.3 (+36.1 points) when the model can use web search, page reading, reverse-image search, and zoom.
- Design philosophy for embodied intelligence: a robot doesn't need every capability in its weights — it needs to recognize when its information is insufficient, pick the right external capability, use it, and continue the task.
More from Embodied
- Luna, a jewelry-style AI wearable that speaks before you ask, launches on Kickstarter — nikola_mr64990 · 2026-11-11
- Dev reverse-engineers Anker AI recording earbuds, ships open-source client app — vista8 · 2026-11-11
- Simulated hand with 39 muscles learns 9/9 poses after 426M training steps — MaziyarPanahi · 2026-11-11
- Versor pitches real2sim2real: robotics sim needs physical dynamics, not just visuals — abhishekunique7 · 2026-11-11
- Origami Robotics and PalatialSim unveil real2sim2real Dexterity Factory, 100 open datasets next week — ybisk · 2026-11-11
- Trainable JWST model boosts exoplanet sensitivity 2–15x in Nature paper — bravo_abad · 2026-11-11