Skip to content
Language models

MiniMax-H3-Image-PE-heretic-NVFP4-GGUF

uncensored quantized version of the MiniMax-H3-Image-PE model

1 min read

This model is a GGUF version of the uncensored narzh49/MiniMax-H3-Image-PE repository, created using the Heretic tool. Quantization is done in the NVFP4 format to optimize memory usage. The base MiniMaxAI/MiniMax-H3-Image-PE model underwent obliteration (removal of built-in censorship restrictions) across layers 25 through 55. This model is useful for researchers and developers who need to work with generative content without the built-in refusals and restrictions of the original version.

Key features

  • GGUF format with NVFP4 quantization to lower hardware requirements
  • Censorship restrictions removed using the Heretic utility via layer modification
  • Row magnitudes preserved (preserve_row_magnitudes: true) during obliteration
  • Lora rank set to 50 for weight adaptation

What you can do

  • Run the model for text generation or multimodal tasks locally with GGUF support
  • Test the model's resistance to refusals using custom prompts
  • Use in text processing pipelines where freedom from the base version's built-in safety filters is required

Example

The model loads through standard tools for the GGUF format and is supported by local backends for running large language models.

Read moreDetails
Training metrics and parameters:
KL divergence (harmless_alpaca): 0.332
Refusals & Evasions (Refusal-Classifier-v2): 34/100 (original is 95/100)
Obliteration affected layers 25 through 55 with lora_rank 50
Discuss Telegram 1 9

Will it run?

Runs in GGUF NVFP4

The most precise of the ready quantised files fits.

  • GGUF NVFP4weights 16 GB · ready files18 GB

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: GGUF NVFP4 — from RTX 4090

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # narzh49/MiniMax-H3-Image-PE-heretic-NVFP4-GGUF
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download narzh49/MiniMax-H3-Image-PE-heretic-NVFP4-GGUF --local-dir ~/models/MiniMax-H3-Image-PE-heretic-NVFP4-GGUF
    # The model is in ~/models. How to run it is in the model card (link above).
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

This is a text modelIt is not the image generator itself but a helper language model (prompt enhancer): it rewrites your short request into a detailed prompt for the generator. That is why the speed-ups here are the LLM ones; the image generator is sped up separately.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

  • MTP (multi-token prediction)0.34.0from November 16, 2026new release

    1.5–2× generation speed

    If the model has MTP layers (DeepSeek-V3/R1, Qwen3-Next, GLM, MiMo and others), it predicts several tokens per step itself.

    How to enable: vLLM / SGLang: --speculative-config '{"method": "mtp", "num_speculative_tokens": 1}'. Check the model card.

    Earlier generations (2)
    • EAGLE-3 — Speculative decoding with a small draft head. For models without their own MTP layers.
    • EAGLE-2 / Medusa — Earlier draft-head generations. Still work if that is all you find.
  • FP8 KV cache0.34.0from November 16, 2026new release

    2× more context in the same memory

    Keeps the attention cache in 8 bits, so long contexts fit where they did not.

    How to enable: vLLM: --kv-cache-dtype fp8; llama.cpp: -ctk q8_0 -ctv q8_0.

  • vLLM / SGLang0.34.0from November 16, 2026new release

    many times more tokens/s on a server

    Server engines: batching, FlashAttention, FP8 and speculative decoding out of the box.

    How to enable: vllm serve <model> — an OpenAI-compatible API.

    Earlier generations (1)
    • llama.cpp — GGUF models on any GPU, Mac or even CPU; some layers can stay in RAM.

MiniMax H3 prompts

All prompts

Ready prompts with the resulting video: copy and try.

Model family

  1. Qwen3.8-27Bfamily root
  2. MiniMax-H3-Image-PE
  3. MiniMax-H3-Image-PE-heretic-NVFP4
  4. MiniMax-H3-Image-PE-heretic-NVFP4-GGUF

New posts about versions of Qwen3.8-27B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 2All on Hugging Face

About the model

narzh49/MiniMax-H3-Image-PE-heretic-NVFP4-GGUF
Author
narzh49
Category
Language models
Released
Posted

On NeyroLearn

Hugging Face

91

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments