Skip to content
Computer vision

VisAlloc-9B

a multimodal model based on Qwen3.5-9B

2 min readupdated

VisAlloc-9B is a multimodal LLM (a large language model that works with both text and images) built on the Qwen/Qwen3.5-9B family and trained using the VisAlloc visual self-distillation method. It takes a full image and a text prompt as input, delivering accurate answers without cropping or zooming into image regions during inference. This makes it useful for visual question answering (VQA) and fine-grained detail recognition in images.

Key features

  • Full-image inference — processes full images through a standard multimodal interface without extra cropping tools.
  • Paired-view self-distillation — trained with positive and negative crops to calibrate the model's visual attention.
  • Selective token supervision — selectively weights tokens based on sensitivity and teacher distribution divergence.
  • Standard model release — complete set of weights in BF16 safetensors format, along with the config, tokenizer, and processor.

What you can do

  • Run the model locally using the standard transformers library to answer questions about images.
  • Serve the model via vLLM with an OpenAI-compatible API to integrate it into third-party services and bots.
  • Analyze visual content inside automated pipelines with high detail accuracy.

Example

from transformers import AutoModelForImageTextToText, AutoProcessor

model_id = "Kaelyn01/VisAlloc-9B"
model = AutoModelForImageTextToText.from_pretrained(
    model_id, dtype="bfloat16", device_map="auto"
)
processor = AutoProcessor.from_pretrained(model_id)

messages = [{
    "role": "user",
    "content": [
        {"type": "image", "url": "[https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg](https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg)"},
        {"type": "text", "text": "What is the color of the label on the bottle in the background?"},
    ],
}]
inputs = processor.apply_chat_template(
    messages,
    add_generation_prompt=True,
    tokenize=True,
    return_dict=True,
    return_tensors="pt",
).to(model.device)
outputs = model.generate(
    **inputs,
    max_new_tokens=512,
    do_sample=False,
    eos_token_id=processor.tokenizer.eos_token_id,
    pad_token_id=processor.tokenizer.pad_token_id,
)
print(processor.decode(
    outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True
))
Read moreDetails
Base model: Qwen/Qwen3.5-9B (9.41B parameters).
Dataset: VisAlloc-6K (5,795 curated examples).
Weights format: BF16 safetensors.
License: Apache-2.0.
Discuss Telegram 1 26

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 18 GB · ready files20 GB
  • FP8 / INT8recommendedweights 8.8 GB10 GB
  • 4-bitweights 4.9 GB5.9 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2080 Ti · BF16 — 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.

    # Kaelyn01/VisAlloc-9B
    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 Kaelyn01/VisAlloc-9B --local-dir ~/models/VisAlloc-9B
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to VisAlloc-9B
  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. Qwen3.5-9B-Basefamily root
  2. Qwen3.5-9B
  3. VisAlloc-9B

New posts about versions of Qwen3.5-9B-Base (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 6All on Hugging Face
Merges 1All on Hugging Face

About the model

Kaelyn01/VisAlloc-9B
Author
Kaelyn01
Category
Computer vision
Released
Posted

On NeyroLearn

Hugging Face

261

Ask AI about this model

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

Comments

Computer vision

Artsy RLHF

Artsy RLHF