AI library

Source · release

Ollama v0.40.13: LFM3 on Linux and Windows

Author
Ollama
Published
29 October 2026
Added
29 October 2026
Original
https://github.com/ollama/ollama/releases/tag/v0.40.13

Summary

Ollama v0.40.13 (2026-10-29) extends Liquid AI's LFM3 models, including lfm3-audio (text-to-speech) and lfm3-omni, to Linux and Windows. In v0.40.12, two days earlier, they ran on Apple hardware only.

Key points

  • Upgrade Ollama, then ollama run lfm3 (or lfm3-vl, lfm3-audio, lfm3-omni).
  • The release notes are short; the d1 2B decision models built on LFM3 are not mentioned, so their Linux and Windows status is unclear.

Why it matters

The LFM3 small models (1.7B to 2B, 256K context) can now be tried locally on any common desktop OS, not just Macs.