Skip to content
Other

Qwen3.8-27B-SuperAdapter GGUF

quantized version of the multimodal language model

2 min readupdated

This model is a collection of quantized GGUF files (format for running large language models on consumer hardware) based on OliviaRossi/Qwen3.8-27B-SuperAdapter. The builder mradermacher prepared the files using an importance matrix (imatrix), which reduces VRAM (video memory) usage with minimal loss in generation quality. The model supports image processing (vision model), and the mmproj vision files are located in a static repository. It is useful for developers and enthusiasts looking to run a large multimodal LLM (large language model) locally on resource-constrained computers.

Key features

  • A range of quantizations from lightweight (Q2_K) to high-quality (Q6_K and Q4_K_M)
  • Use of imatrix (importance matrix) to optimize compression quality
  • Support for multimodal tasks (vision model for image processing)
  • Optimized for running on local hardware via GGUF-compatible clients

What you can do

  • Run the model locally in LM Studio, Ollama, or llama.cpp for everyday text and visual tasks
  • Choose the optimal quantization (for example, Q4_K_M) for your PC's available VRAM
  • Integrate the multimodal model into local text and image processing pipelines

Example

The process for running and working with GGUF files follows the standard llama.cpp ecosystem. Use your chosen quantized file (such as Q4_K_M) in your favorite GGUF client, following the instructions from classic guides by TheBloke.

Read moreDetails
Base model: OliviaRossi/Qwen3.8-27B-SuperAdapter
Quantization type: imatrix / GGUF (including Q4_K_M, Q6_K, IQ4_XS)
File size: from 11.3 GB (Q2_K) to 22.5 GB (Q6_K)
Features: this is a vision model, mmproj files are in a static repository
Discuss Telegram 5 442

Will it run?

Runs in GGUF Q6_K

The most precise of the ready quantised files fits.

  • GGUF Q6_Kweights 21 GB · ready files23 GB
  • GGUF Q5_K_Mweights 18 GB · ready files21 GB
  • GGUF Q5_K_Sweights 18 GB · ready files20 GB
  • GGUF Q4_1weights 16 GB · ready files18 GB
  • GGUF Q4_K_Mweights 16 GB · ready files18 GB
  • GGUF Q4_K_Sweights 15 GB · ready files17 GB
  • GGUF Q4_0weights 15 GB · ready files17 GB
  • GGUF IQ4_XSweights 14 GB · ready files16 GB
  • GGUF Q3_K_Lweights 14 GB · ready files16 GB
  • GGUF Q3_K_Mweights 13 GB · ready files15 GB
  • GGUF IQ3_Mweights 12 GB · ready files14 GB
  • GGUF IQ3_Sweights 12 GB · ready files14 GB
  • GGUF Q3_K_Sweights 11 GB · ready files13 GB
  • GGUF Q2_Kweights 10 GB · ready files12 GB

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: GGUF Q6_K — 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.

    # mradermacher/Qwen3.8-27B-SuperAdapter-i1-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 mradermacher/Qwen3.8-27B-SuperAdapter-i1-GGUF --local-dir ~/models/Qwen3.8-27B-SuperAdapter-i1-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.

  • 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.

Model family

  1. Qwen3.8-27Bfamily root
  2. Qwen3.8-27B-SuperAdapter
  3. Qwen3.8-27B-SuperAdapter-i1-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

mradermacher/Qwen3.8-27B-SuperAdapter-i1-GGUF
Author
mradermacher
Category
Other
Released
Posted

On NeyroLearn

Hugging Face

4425

Ask AI about this model

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

Comments