Skild AI's S1 model learns new tasks from one video example

Startup Skild AI introduced the S1 model, which learns new tasks from a single video example without fine-tuning. On previously unseen tasks, S1 achieved 66% success versus 9% for language-based VLA models, with setup taking minutes instead of 50–100 hours of teleoperation.

• S1 achieved 66% success on new tasks vs 9% for VLA models
• One video example replaces about 380 training examples
• Collecting 380 examples requires 50–100 hours of teleoperation
• In-context advantage appears only with large data volumes