SkildAI unveils S1, a foundation model that learns 10-minute tasks from one video example
deepakpathak · x · 2026-08-27
SkildAI has introduced S1, a new foundation model designed for one-shot learning. S1 can be taught tasks it has never seen before, lasting up to 10 minutes, using just a single video prompt without any fine-tuning. A live demonstration shows S1 operating in real-time via in-context learning, highlighting significant advancements in embodied intelligence and task generalization capabilities.
More from Embodied
- Robot Figure Skating Demo Sparks Call for Robo-Olympics in France — IgorCarron · 2026-08-27
- Polymarket gives Tesla just 17% odds of launching California robotaxis by year-end — Polymarket · 2026-08-27
- Fei-Fei Li's World Labs hires Business Lead for Robotics & Physical AI — theworldlabs · 2026-08-27
- Ludi 0.1: Agentic System for Socially Intelligent Robots — rdnowak · 2026-08-27
- Skeptical of comparing egocentric robot data collection to Tesla FSD — binarybits · 2026-08-27
- Weekly Recap: Robot records, XPeng funding, and A2A protocol governance — Dapper-Tale-4021 · 2026-08-27