What are the hardware requirements for MiniMax H3 Image PE NVFP4 MTP Repacked?

MiniMax H3 Image PE NVFP4 MTP Repacked requires approximately 18.8 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed MiniMax H3 Image PE NVFP4 MTP Repacked and how large is it?

MiniMax H3 Image PE NVFP4 MTP Repacked is developed by prithivMLmods, a model with 27.8b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate MiniMax H3 Image PE NVFP4 MTP Repacked?

MiniMax H3 Image PE NVFP4 MTP Repacked is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

MiniMax H3 Image PE NVFP4 MTP Repacked by prithivMLmods — VRAM 18.8GB

Name: MiniMax H3 Image PE NVFP4 MTP Repacked
Author: prithivMLmods

MiniMax H3 Image PE NVFP4 MTP Repacked is an open-source language model by prithivMLmods. Features: 27.8b LLM, VRAM: 18.8GB, License: other.

27b 8-bit B200 B300 Base model:minimaxai/minimax-h... Base model:quantized:minimaxai... Blackwell Conversational En Endpoints compatible Fp4 Gb202 Image-text-to-text Llama-cli Llama-cpp Minimax-h3 Model optimizer Modelopt Mtp Nvfp4 Nvidia Nvidia-blackwell Prompt-expansion Qwen3.5 Qwen3.6 Qwen3.8 Qwen3 5 Region:us Rtx-pro-6000 Safetensors Sglang Sharded Sm120 Speculative-decoding Stable Tensorflow Text-to-image

Model Card on HF 🤗: https://huggingface.co/prithivMLmods/MiniMax-H3-Image-PE-NVFP4-MTP-Repacked

MiniMax H3 Image PE NVFP4 MTP Repacked Parameters and Internals

LLM Name	MiniMax H3 Image PE NVFP4 MTP Repacked
Repository 🤗	https://huggingface.co/prithivMLmods/MiniMax-H3-Image-PE-NVFP4-MTP-Repacked
Base Model(s)	MiniMax H3 Image PE MiniMaxAI/MiniMax-H3-Image-PE
Model Size	27.8b
Required VRAM	18.8 GB
Updated	2026-11-14
Maintainer	prithivMLmods
Model Type	qwen3_5
Model Files	5.2 GB: 1-of-4 5.2 GB: 2-of-4 5.2 GB: 3-of-4 3.2 GB: 4-of-4
Supported Languages	en
Model Architecture	Qwen3_5ForConditionalGeneration
License	other
Model Max Length	262144
Transformers Version	5.14.1
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Errors	replace

Rank the MiniMax H3 Image PE NVFP4 MTP Repacked capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to MiniMax H3 Image PE NVFP4 MTP Repacked

Best Alternatives	Context / RAM	Downloads	Likes
Feyospace V1.1 Small	0K / 55.6 GB	1038	11
Feyospace S1 Obliterated	0K / 109.3 GB	392	5
StarLife V1 WSV	0K / 55.6 GB	81	1
Feyospace V1.1 Small NVFP4	0K / 23 GB	14	1
Hemmingway 1 VL	0K / 55.4 GB	22	0
... Obliterated OQ8e FP16 MTP MLX	0K / 30.9 GB	155	2
... Obliterated OQ4e FP16 MTP MLX	0K / 17.9 GB	142	2
... Obliterated OQ6e FP16 MTP MLX	0K / 24.5 GB	101	2
Feyospace V1.1 Small FP8	0K / 30.6 GB	118	2

Note: green Score (e.g. "73.2") means that the model is better than prithivMLmods/MiniMax-H3-Image-PE-NVFP4-MTP-Repacked.