---
type: "post"
id: "deepbeepmeep-publishes-a-quantized-repack-of-minimax-h3-d7ffe5"
slug: "deepbeepmeep-publishes-a-quantized-repack-of-minimax-h3-d7ffe5"
title: "DeepBeepMeep publishes a quantized repack of MiniMax H3 for GPU-poor local inference via WanGP."
description: "DeepBeepMeep has published a Hugging Face repository, MiniMax-H3, containing MiniMax H3 image-model files intended for use with WanGP, its local web interface for generative models."
retrieval_nugget: "DeepBeepMeep has published a Hugging Face repository, MiniMax-H3, containing MiniMax H3 image-model files intended for use with WanGP, its local web interface for generative models."
published_at: "2026-10-25"
updated_at: "2026-10-25"
record_date: "2026-10-25"
date_kind: "published_at"
topics: ["generative-media","multimodal"]
entities: ["DeepBeepMeep","MiniMax H3","WanGP","Hugging Face","LightVAE Lite","TaoMate H3 FL2VA 3-Step"]
related_patterns: ["generative-media-becomes-application-infrastructure"]
novelty_label: "new_fact"
what_changed: "DeepBeepMeep adds a quantized MiniMax H3 repack for WanGP, including a rank-128 adapter and BF16/INT8 LightVAE Lite decoder options."
prior_refs: ["memory:26e764cc-5941-84fd-aae6-7f7367180c8f"]
source_url: "https://huggingface.co/DeepBeepMeep/MiniMax-H3"
source_urls: ["https://huggingface.co/DeepBeepMeep/MiniMax-H3"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:deepbeepmeep-publishes-a-quantized-repack-of-minimax-h3-d7ffe5"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/deepbeepmeep-publishes-a-quantized-repack-of-minimax-h3-d7ffe5/","markdown":"https://newruntime.com/posts/deepbeepmeep-publishes-a-quantized-repack-of-minimax-h3-d7ffe5.md","json":"https://newruntime.com/posts/deepbeepmeep-publishes-a-quantized-repack-of-minimax-h3-d7ffe5.json"}
---

# DeepBeepMeep publishes a quantized repack of MiniMax H3 for GPU-poor local inference via WanGP.

## Retrieval answer

DeepBeepMeep has published a Hugging Face repository, MiniMax-H3, containing MiniMax H3 image-model files intended for use with WanGP, its local web interface for generative models.

DeepBeepMeep has published a Hugging Face repository, MiniMax-H3, containing MiniMax H3 image-model files intended for use with WanGP, its local web interface for generative models. The repository is labelled as a 26B-parameter qwen3vl architecture and includes GGUF variants.

The model page lists a Q2_K variant at 8.49 GB and a Q4_K_M variant at 14.6 GB. It says WanGP supports Wan-derived models, Hunyuan Video, Flux, Qwen, Z-Image, and LTX models; its stated features include automatic model downloads matched to a user's architecture. The page also says some WanGP-supported models can run with as little as 6 GB of VRAM, and that older RTX 10XX and 20xx GPUs are supported.

An October 15, 2026 update lists a rank-128 adapter for the TaoMate H3 FL2VA 3-Step profile and LightVAE Lite decoder files in BF16 and INT8 ConvRot. The source does not provide benchmark results, measured H3 memory use, generation quality comparisons, or a complete account of licence terms for every included component.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Generative Media](https://newruntime.com/topics/generative-media.md): Explore the generative-media topic hub.
- topic hub: [Multimodal](https://newruntime.com/topics/multimodal.md): Explore the multimodal topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)