LiquidAI d1-2B decision model runs fully local on M5 Max: ~14 decisions/s, zero tokens

iamrobotbear · x · 2026-11-17

Developer Ivan Fioravanti demoed LiquidAI's d1-2B decision model via mlx-vlm on an Apple Silicon M5 Max MacBook Pro: it reads gestures and expressions from a webcam in 68 ms, makes 14 decisions per second, generates zero tokens, and runs 100% locally — a showcase of small decision models for real-time on-device perception.

Related event: LiquidAI d1-2B Runs Locally: 14 Decisions Per Second With Zero Tokens(2 posts)→

Original post →

More from Infra

Infra channel →