Civitai finally adds MiniMax H3 Image filter, but LoRA ecosystem stays quiet

collectedtooth404 · reddit · 2026-11-19

Civitai has finally added a search filter for MiniMax H3 Image LoRAs and custom checkpoints — image generations had the filter since launch, but finding models by it wasn't possible until now, likely due to a bug.

The author also notes surprisingly few LoRAs and shared resources for H3 Image: initial fears that it was hard to train proved unfounded, as training with Fizgig worked fine. The fix means users no longer need to scour Hugging Face or torrents for H3 LoRAs.

Original post →

More from Multimodal

Multimodal channel →