Speech Models

Reporting on Speech Models from Lazy Tensor.

Search this sectionA place, person or subject

Recursive Velocity and Physical Models

Anthropic doubles development speed as startups target edge AI

Independent evaluators at METR find Anthropic is improving model capabilities twice as fast as a year ago while Physicl raises $10 million and Underdog releases a 32MB voice model.

Nov 10, 2026By Lazy Tensor3 min read

Conceptual editorial illustration depicting robotic simulation calibration, recursive model evaluation, and compact edge acoustic processing.

Berkeley

Speech Models Edge AI Anthropic