Qwen3-VL-235B-A22B-Thinking-NVFP4

Model Overview

• Model Architecture: Qwen/Qwen3-VL-235B-A22B-Thinking
• Model Optimizations:
  • Weight quantization: FP4
  • Activation quantization: FP4
• Out-of-scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in languages other than English.
• Release Date: 12/05/2025
• Version: 1.0
• Model Developers: GaleneAI

This model is a quantized version of Qwen/Qwen3-VL-235B-A22B-Thinking. It was evaluated on a several tasks to assess the its quality in comparison to the unquatized model.

Model Optimizations

This model was obtained by quantizing the weights and activations of Qwen/Qwen3-VL-235B-A22B-Thinking to FP4 data type, ready for inference with vLLM>=0.11.0 This optimization reduces the number of bits per parameter from 16 to 4, reducing the disk size and GPU memory requirements by approximately 75%.

Only the weights and activations of the linear operators within transformers blocks are quantized using LLM Compressor.

Deployment

Use with vLLM

This model can be deployed efficiently using the vLLM backend, as shown in the example below.

from vllm import LLM, SamplingParams
from transformers import AutoTokenizer

model_id = "GaleneAI/Qwen3-VL-235B-A22B-Thinking-NVFP4"
number_gpus = 1

sampling_params = SamplingParams(temperature=0.6, top_p=0.9)

tokenizer = AutoTokenizer.from_pretrained(model_id)

messages = [
    {"role": "system", "content": "You are a proficient Rust developer."},
    {"role": "user", "content": "Provide me with a quicksort implementation in Rust."},
]

prompts = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False)

llm = LLM(model=model_id, tensor_parallel_size=number_gpus)

outputs = llm.generate(prompts, sampling_params)

generated_text = outputs[0].outputs[0].text
print(generated_text)

vLLM aslo supports OpenAI-compatible serving. See the documentation for more details.

Creation

This model was created by applying LLM Compressor with calibration samples from UltraChat, as presented in the code snipet below.

import torch
from datasets import load_dataset
from transformers import AutoProcessor, Qwen3VLMoeForConditionalGeneration

from llmcompressor import oneshot
from llmcompressor.modeling import replace_modules_for_calibration
from llmcompressor.modifiers.quantization import QuantizationModifier
from llmcompressor.utils import dispatch_for_generation

# NOTE: Requires a minimum of transformers 4.57.0

MODEL_ID = "Qwen/Qwen3-VL-235B-A22B-Thinking"


# Load model.
model = Qwen3VLMoeForConditionalGeneration.from_pretrained(MODEL_ID, torch_dtype="auto")
processor = AutoProcessor.from_pretrained(MODEL_ID)
model = replace_modules_for_calibration(model)

DATASET_ID = "neuralmagic/calibration"
NUM_CALIBRATION_SAMPLES = 20
MAX_SEQUENCE_LENGTH = 8192

ds = load_dataset(DATASET_ID, name="LLM", split=f"train[:{NUM_CALIBRATION_SAMPLES}]")


def preprocess_function(example):
    messgages = []
    for message in example["messages"]:
        messgages.append(
            {
                "role": message["role"],
                "content": [{"type": "text", "text": message["content"]}],
            }
        )

    return processor.apply_chat_template(
        messgages,
        return_tensors="pt",
        padding=False,
        truncation=True,
        max_length=MAX_SEQUENCE_LENGTH,
        tokenize=True,
        add_special_tokens=False,
        return_dict=True,
        add_generation_prompt=False,
    )


ds = ds.map(preprocess_function, batched=False, remove_columns=ds.column_names)


def data_collator(batch):
    assert len(batch) == 1
    return {
        key: (
            torch.tensor(value)
            if key != "pixel_values"
            else torch.tensor(value, dtype=torch.bfloat16).squeeze(0)
        )
        for key, value in batch[0].items()
    }


# Configure the quantization algorithm and scheme.
# In this case, we:
#   * quantize the weights to fp4 with group-wise quantization
#   * quantize the activations to fp4 with dynamic group activations
recipe = QuantizationModifier(
    targets="Linear",
    scheme="NVFP4",
    ignore=[
        "re:.*lm_head",
        "re:visual.*",
        "re:model.visual.*",
        "re:.*mlp.gate$",
    ],
)

# Apply quantization.
oneshot(
    model=model,
    recipe=recipe,
    max_seq_length=MAX_SEQUENCE_LENGTH,
    num_calibration_samples=NUM_CALIBRATION_SAMPLES,
    dataset=ds,
    data_collator=data_collator,
)

# Save to disk in compressed-tensors format.
SAVE_DIR = MODEL_ID.rstrip("/").split("/")[-1] + "-NVFP4"
model.save_pretrained(SAVE_DIR)
processor.save_pretrained(SAVE_DIR)

print("========== SAMPLE GENERATION ==============")
dispatch_for_generation(model)
input_ids = processor(text="Hello my name is", return_tensors="pt").input_ids.to("cuda")
output = model.generate(input_ids, max_new_tokens=20)
print(processor.decode(output[0]))
print("==========================================")

✨ Maintained by GaleneAI