Dev strips decoding from Qwen 3.5 9B to serve token probabilities locally at 100ms

telltaleatheist · reddit · 2026-09-23

A Reddit user shared "jev", a local tool that cuts off the decoding stage of qwen 3.5:9b, retrieves the raw softmax output, and returns token probabilities as a structured JSON payload. The author reports 100ms responses running fully locally for free, and notes the vision tower remains usable for image analysis. It targets use cases that need probability distributions rather than generated text.

Original post →

More from Models

Models channel →