🎬 Video · Free

MiniMax H3 Turbo LoRAs (4–8 Step)

A follow-up to my MiniMax H3 workflow: there are now two turbo LoRAs that cut sampling from ~20 steps down to 4-8 — roughly a 4x speedup — while keeping H3's native audio. This post covers both, with attached workflows for each and a straight recommendation on which to use. If you haven't set up H3 yet, start with the full workflow and memory guide first:

✓ 100% free — download the workflow and run it locally.
⬇ MiniMax-H3-Speed-up-LightX2V-4-8-steps-Turbo ⬇ MiniMax-H3-Speed-up-Larryvrh-4-8-steps-Turbo

Requirements

Model downloads

Custom nodes

Setup steps

  1. Steps & Strength (Applies to Both)
  2. 4 steps — fast, and genuinely usable on the Larryvrh LoRA. In my testing the Larryvrh version held sharp detail and clean audio even at 4. On the lightx2v LoRA, 4 is more of a fast-draft mode (softer detail).
  3. 6 steps — the safe balanced default for either LoRA. Full speed benefit, solid detail, clean audio.
  4. 8 steps — go here if you want the cleanest possible audio. At the very lowest step counts the native audio degrades first, so bump to 8 when audio quality is critical.

Notes & tips

More resources

Want the 1-click version?

Skip the manual model downloads and node installs — get a ready-to-run installer, or unlock all 75+ with Local Lab Pro.