Instructions to use amd/Qwen3.8-Flash-Next-Quark-MXFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use amd/Qwen3.8-Flash-Next-Quark-MXFP4 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="amd/Qwen3.8-Flash-Next-Quark-MXFP4")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("amd/Qwen3.8-Flash-Next-Quark-MXFP4")
model = AutoModelForMultimodalLM.from_pretrained("amd/Qwen3.8-Flash-Next-Quark-MXFP4", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use amd/Qwen3.8-Flash-Next-Quark-MXFP4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "amd/Qwen3.8-Flash-Next-Quark-MXFP4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "amd/Qwen3.8-Flash-Next-Quark-MXFP4",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/amd/Qwen3.8-Flash-Next-Quark-MXFP4

SGLang

How to use amd/Qwen3.8-Flash-Next-Quark-MXFP4 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "amd/Qwen3.8-Flash-Next-Quark-MXFP4" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "amd/Qwen3.8-Flash-Next-Quark-MXFP4",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "amd/Qwen3.8-Flash-Next-Quark-MXFP4" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "amd/Qwen3.8-Flash-Next-Quark-MXFP4",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use amd/Qwen3.8-Flash-Next-Quark-MXFP4 with Docker Model Runner:
```
docker model run hf.co/amd/Qwen3.8-Flash-Next-Quark-MXFP4
```

Model Overview

Model Architecture: Qwen4ExpForConditionalGeneration
- Input: Text, Image, Video
- Output: Text
Supported Hardware Microarchitecture: AMD MI350 / MI355
ROCm: 7.2.3
PyTorch: 2.12.0+git6bbd260
Transformers: 5.16.1
Operating System(s): Linux
Inference Engine: vLLM
Model Optimizer: AMD-Quark
- Quantized layers: All MoE experts in the language model, including the shared expert. Attention is kept in bf16.
- Weight quantization:
  - MOE OCP MXFP4, Static
- Activation quantization:
  - MOE OCP MXFP4, Dynamic

This checkpoint quantizes the routed and shared MoE experts to OCP MXFP4 and keeps everything else in bf16, including the full self-attention and the (Mamba-style) linear-attention projections, the MTP head, router gates, shared-expert gate, PLE n-gram path, vision tower, hyper-connection mixers, and lm_head.

Model Quantization

The model was quantized from Qwen3.8-Flash-Next using AMD-Quark. MoE expert weights and activations are quantized to OCP MXFP4.

Quantization script:

from quark.torch import LLMTemplate, ModelQuantizer
from quark.torch.quantization.weight_convert import SplitFusedExperts, WeightConverter

EXCLUDE = [
    "lm_head", "*mlp.gate", "*shared_expert_gate", "mtp.*", "model.visual.*",
    "*hyper_connection*", "*ple.*", "*linear_attn.*", "*self_attn.*",
]

template = LLMTemplate(
    model_type="qwen4_exp",
    kv_layers_name=["*self_attn.k_proj", "*self_attn.v_proj"],
    q_layer_name="*self_attn.q_proj",
    exclude_layers_name=EXCLUDE,
    f2f_weight_converters=[
        WeightConverter("gate_up_proj", ["gate_proj.weight", "up_proj.weight"],
                        operations=[SplitFusedExperts(split_axis=0)]),
        WeightConverter("down_proj", ["down_proj.weight"],
                        operations=[SplitFusedExperts(split_axis=0)]),
    ],
)
LLMTemplate.register_template(template)

quant_config = template.get_config(scheme="mxfp4", exclude_layers=EXCLUDE)

quantizer = ModelQuantizer(quant_config)
quantizer.direct_quantize_checkpoint(
    pretrained_model_path="Qwen/Qwen3.8-Flash-Next",
    save_path=OUTPUT_DIR,
    weight_converters=template.f2f_weight_converters,
    device="cuda",
)

For further details or issues, please refer to the AMD-Quark documentation.

Evaluation

The model was evaluated using vLLM, EvalScope, and lm-evaluation-harness.

Accuracy

Benchmark	Qwen/Qwen3.8-Flash-Next	amd/Qwen3.8-Flash-Next-Quark-MXFP4 (this model)	Recovery
GSM8K (5-shot)	96.8	96.5	99.7%
GPQA Diamond (mean@8)	91.67	91.67	100.00%
AIME26 (mean@16)	99.17	98.33	99.15%
MMMU Pro, standard (mean@1)	79.65	80.00	100.44%
MMMU Pro, vision (mean@1)	78.38	77.40	98.75%
SciCode, subproblem (mean@8)	49.01	48.58	99.12%
AA-LCR (mean@16)	78.50	80.63	102.71%
IFBench (mean@5)	83.27	82.67	99.28%

Note: Results may vary from run to run.

Reproduction

The following results were obtained using vLLM and EvalScope 1.14.1. SciCode uses sandbox image scicode-benchmark:7687150b169df5091c158abb.

Launching server

We launch the vLLM server following vLLM Recipes using container image vllm/vllm-openai-rocm:nightly-83c078d955cd6e0c63d3b12abe669c8de936a370.

VLLM_USE_RUST_FRONTEND=1 \
VLLM_ROCM_USE_AITER=1 \
VLLM_ROCM_USE_AITER_FUSION_SHARED_EXPERTS=1 \
AITER_USE_FLYDSL_MOE_SORTING=1 \
vllm serve amd/Qwen3.8-Flash-Next-Quark-MXFP4 \
  --host 0.0.0.0 \
  --port 8000 \
  --tensor-parallel-size 2 \
  --gdn-prefill-backend aiter_flydsl \
  --gpu-memory-utilization 0.90 \
  --max-num-seqs 256 \
  --enable-prefix-caching \
  --enable-auto-tool-choice \
  --tool-call-parser qwen3_coder \
  --reasoning-parser qwen3

Evaluating model in a new terminal

GSM8K

lm_eval --model local-chat-completions \
    --model_args "model=amd/Qwen3.8-Flash-Next-Quark-MXFP4,base_url=http://localhost:8000/v1/chat/completions,num_concurrent=32,timeout=1200,max_retries=3,tokenized_requests=False,max_gen_toks=2048" \
    --tasks gsm8k --num_fewshot 5 --apply_chat_template

GPQA Diamond

evalscope eval \
  --model amd/Qwen3.8-Flash-Next-Quark-MXFP4 \
  --api-url http://127.0.0.1:8000/v1/chat/completions \
  --datasets gpqa_diamond \
  --eval-batch-size 8 \
  --repeats 8 \
  --generation-config '{"temperature":1.0,"top_p":0.95,"max_tokens":131072,"timeout":7200}' \
  --work-dir logs/gpqa_diamond

AIME26

evalscope eval \
  --model amd/Qwen3.8-Flash-Next-Quark-MXFP4 \
  --api-url http://127.0.0.1:8000/v1/chat/completions \
  --datasets aime26 \
  --eval-batch-size 8 \
  --repeats 16 \
  --generation-config '{"temperature":1.0,"top_p":0.95,"max_tokens":131072,"timeout":7200}' \
  --work-dir logs/aime26

MMMU Pro (standard & vision)

evalscope eval \
  --model amd/Qwen3.8-Flash-Next-Quark-MXFP4 \
  --api-url http://127.0.0.1:8000/v1/chat/completions \
  --datasets mmmu_pro \
  --eval-batch-size 8 \
  --repeats 1 \
  --generation-config '{"temperature":1.0,"top_p":0.95,"max_tokens":98304,"timeout":7200}' \
  --dataset-args '{"mmmu_pro":{"dataset_format":"standard (10 options)"}}' \
  --work-dir logs/mmmu_pro_standard

evalscope eval \
  --model amd/Qwen3.8-Flash-Next-Quark-MXFP4 \
  --api-url http://127.0.0.1:8000/v1/chat/completions \
  --datasets mmmu_pro \
  --eval-batch-size 8 \
  --repeats 1 \
  --generation-config '{"temperature":1.0,"top_p":0.95,"max_tokens":98304,"timeout":7200}' \
  --dataset-args '{"mmmu_pro":{"dataset_format":"vision"}}' \
  --work-dir logs/mmmu_pro_vision

SciCode

evalscope eval \
  --model amd/Qwen3.8-Flash-Next-Quark-MXFP4 \
  --api-url http://127.0.0.1:8000/v1/chat/completions \
  --datasets scicode \
  --dataset-hub modelscope \
  --eval-batch-size 8 \
  --repeats 8 \
  --generation-config '{"temperature":1.0,"top_p":0.95,"max_tokens":131072,"timeout":7200}' \
  --dataset-args '{"scicode":{"extra_params":{"provide_background":true}}}' \
  --sandbox '{"enabled":true,"engine":"docker","pool_size":4}' \
  --work-dir logs/scicode

AA-LCR (v1.1)

Judge configuration (aa-lcr-judge.json; replace <YOUR_SUBSCRIPTION_KEY>):

{
  "strategy": "llm",
  "models": [
    {
      "model_id": "gpt-5.6-luna",
      "api_url": "https://llm-api.amd.com/openai/responses",
      "api_key": "EMPTY",
      "eval_type": "openai_responses_api",
      "generation_config": {
        "reasoning_effort": "medium",
        "max_tokens": 4096,
        "timeout": 7200,
        "extra_query": {
          "api-version": "2025-04-01-preview"
        },
        "extra_headers": {
          "Ocp-Apim-Subscription-Key": "<YOUR_SUBSCRIPTION_KEY>"
        }
      }
    }
  ]
}

evalscope eval \
  --model amd/Qwen3.8-Flash-Next-Quark-MXFP4 \
  --api-url http://127.0.0.1:8000/v1/chat/completions \
  --datasets aa_lcr \
  --dataset-hub huggingface \
  --eval-batch-size 8 \
  --repeats 16 \
  --generation-config '{"temperature":1.0,"top_p":0.95,"max_tokens":131072,"timeout":7200}' \
  --dataset-args '{"aa_lcr":{"hf_dataset_revision":"9a77ef56b717057ade24ceab4d273712a0b4f19e"}}' \
  --judge "$(cat aa-lcr-judge.json)" \
  --work-dir logs/aa_lcr

IFBench

evalscope eval \
  --model amd/Qwen3.8-Flash-Next-Quark-MXFP4 \
  --api-url http://127.0.0.1:8000/v1/chat/completions \
  --datasets ifbench \
  --dataset-hub huggingface \
  --eval-batch-size 8 \
  --repeats 5 \
  --generation-config '{"temperature":1.0,"top_p":0.95,"max_tokens":131072,"timeout":7200}' \
  --dataset-args '{"ifbench":{"primary_metric":"prompt_level_loose"}}' \
  --work-dir logs/ifbench

License

Downloads last month: 14,046

Safetensors

Model size

119B params

Tensor type

BF16

I64

Model tree for amd/Qwen3.8-Flash-Next-Quark-MXFP4

Base model

Qwen/Qwen3.8-Flash-Next

Quantized

(586)

this model

Quantizations

1 model