Instructions to use AxionML/DeepSeek-V4-Flash-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AxionML/DeepSeek-V4-Flash-NVFP4 with Transformers:

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

pipe = pipeline("text-generation", model="AxionML/DeepSeek-V4-Flash-NVFP4")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("AxionML/DeepSeek-V4-Flash-NVFP4")
model = AutoModelForCausalLM.from_pretrained("AxionML/DeepSeek-V4-Flash-NVFP4", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use AxionML/DeepSeek-V4-Flash-NVFP4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AxionML/DeepSeek-V4-Flash-NVFP4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AxionML/DeepSeek-V4-Flash-NVFP4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/AxionML/DeepSeek-V4-Flash-NVFP4

SGLang

How to use AxionML/DeepSeek-V4-Flash-NVFP4 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 "AxionML/DeepSeek-V4-Flash-NVFP4" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AxionML/DeepSeek-V4-Flash-NVFP4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "AxionML/DeepSeek-V4-Flash-NVFP4" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AxionML/DeepSeek-V4-Flash-NVFP4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use AxionML/DeepSeek-V4-Flash-NVFP4 with Docker Model Runner:
```
docker model run hf.co/AxionML/DeepSeek-V4-Flash-NVFP4
```

AxionML DeepSeek-V4-Flash-NVFP4

Mirrored by AxionML for open-source serving and deployment use cases. Part of AxionML's effort to provide ready-to-serve quantized models for the community.

Quantized by NVIDIA. The weights in this repository are an unmodified copy of nvidia/DeepSeek-V4-Flash-NVFP4 (revision e3cd60e7de98e9867116860d522499a728de1cf9). All credit for the quantization belongs to NVIDIA.

This is an NVFP4-quantized version of deepseek-ai/DeepSeek-V4-Flash (284B total parameters, 13B activated), quantized with NVIDIA Model Optimizer.

About NVFP4 quantization: NVFP4 on Blackwell couples a compact E2M1 FP4 codebook with blockwise FP8 (E4M3) scaling over 16-element micro-blocks, so that 4-bit stored values remain numerically useful for neural-network computation. The E2M1 codebook provides a small, nonuniform set of representable magnitudes up to ±6 and relies on saturating behavior rather than IEEE NaN/Inf encodings to maximize usable range per bit. Using an FP8 block scale (rather than power-of-two-only E8M0) enables fractional scales and error-minimizing scale selection. On Blackwell Tensor Cores, native FP4 multipliers exploit E2M1 simplicity while higher-precision FP32 accumulation protects dot-product accuracy.

Ready for commercial and non-commercial use under MIT.

Model Summary


Architecture	MoE with hybrid attention (Compressed Sparse + Heavily Compressed Attention), Manifold-Constrained Hyper-Connections
Total Parameters	284B
Activated Parameters	13B
Reasoning Modes	Non-think / Think High / Think Max
Context Length	1M tokens
Checkpoint Size	~168 GB

Evaluation Results

Benchmark	Baseline	NVFP4
GPQA Diamond	89.4	89.1
AA-LCR	65.8	65.5
τ²-Bench Telecom	94.3	94.2
SciCode	48.1	48.1
IFBench	78.8	79.5

Scores reported by NVIDIA for this checkpoint. temperature=1.0, top_p=1.0, max 384,000 tokens.

Quantization Details

Quantization format: NVFP4 weights and activations for the routed MoE experts; attention and shared experts keep their source FP8 precision
Calibration dataset: cnn_dailymail, Nemotron-Post-Training-Dataset-v2
Tool: NVIDIA Model Optimizer v0.44.0

Usage

Deploy with SGLang

python3 -m sglang.launch_server \
    --model-path AxionML/DeepSeek-V4-Flash-NVFP4 \
    --tensor-parallel-size 8 \
    --trust-remote-code

Deploy with vLLM

vllm serve AxionML/DeepSeek-V4-Flash-NVFP4 \
    --tensor-parallel-size 4 \
    --trust-remote-code \
    --kv-cache-dtype fp8

SGLang auto-detects NVFP4 from hf_quant_config.json (sgl-project/sglang#25820). vLLM validated upstream on GB300 (TP4). Superseded by AxionML/DeepSeek-V4-Flash-0731-NVFP4.

Limitations

The base model was trained on data that may contain toxic language and societal biases. The quantized model inherits these limitations. It may generate inaccurate, biased, or offensive content. Please refer to the original model card and the upstream quantized model card for full details.

Credits

Base model: deepseek-ai/DeepSeek-V4-Flash
Quantization: nvidia/DeepSeek-V4-Flash-NVFP4 by NVIDIA
Mirror: AxionML

Downloads last month: 168

Safetensors

Model size

291B params

Tensor type

BF16

F8_E4M3

I64

F32

Model tree for AxionML/DeepSeek-V4-Flash-NVFP4

Base model

deepseek-ai/DeepSeek-V4-Flash

Quantized

(127)

this model

AxionML
/

DeepSeek-V4-Flash-NVFP4