deepseek-ai
/

DeepSeek-V4-Flash-0731

Text Generation

8-bit precision

Model card Files Files and versions

Instructions to use deepseek-ai/DeepSeek-V4-Flash-0731 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use deepseek-ai/DeepSeek-V4-Flash-0731 with Transformers:

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

pipe = pipeline("text-generation", model="deepseek-ai/DeepSeek-V4-Flash-0731")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("deepseek-ai/DeepSeek-V4-Flash-0731")
model = AutoModelForCausalLM.from_pretrained("deepseek-ai/DeepSeek-V4-Flash-0731", device_map="auto")

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use deepseek-ai/DeepSeek-V4-Flash-0731 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "deepseek-ai/DeepSeek-V4-Flash-0731"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "deepseek-ai/DeepSeek-V4-Flash-0731",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/deepseek-ai/DeepSeek-V4-Flash-0731

How to use deepseek-ai/DeepSeek-V4-Flash-0731 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 "deepseek-ai/DeepSeek-V4-Flash-0731" \
    --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": "deepseek-ai/DeepSeek-V4-Flash-0731",
		"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 "deepseek-ai/DeepSeek-V4-Flash-0731" \
        --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": "deepseek-ai/DeepSeek-V4-Flash-0731",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use deepseek-ai/DeepSeek-V4-Flash-0731 with Docker Model Runner:
```
docker model run hf.co/deepseek-ai/DeepSeek-V4-Flash-0731
```

Resources

View closed (7)

Add Toolathlon-Verified evaluation result

#75 opened 17 days ago by

Deepseek V4 Flash Lite Idea

#74 opened 18 days ago by

Urgent matter requiring a meeting with the Grandmaster.

#73 opened 22 days ago by

Sick model, deep sick! 😍

#72 opened 24 days ago by

Speed benchmarking 2xSpark vs 2xRTX Pro 6000

#71 opened 24 days ago by

whistlercapital

Can i run it?

#70 opened 25 days ago by

Add Terminal-Bench evaluation results

#69 opened 28 days ago by

Papamoth

#68 opened 29 days ago by

vx大模型技术交流

#67 opened 30 days ago by

Community implementation: native DeepSeek-V4-Flash-0731 inference in pure C on CPU

#66 opened about 1 month ago by

Grabage token looping issue

#65 opened about 1 month ago by

Issue running this model

#64 opened about 1 month ago by

nandakishore071

Hermes-Agent and Silly Tavern speed Improvements by preventing context corruption

#62 opened about 1 month ago by

why models output or use wrong absolute path.

#61 opened about 1 month ago by

Degenerate repetition at tool-call decision points in long agentic contexts (portable reproducer; 3 serving stacks incl. Ollama cloud 6/6)

#58 opened about 1 month ago by

We threw 256 prompt-injection attacks at DeepSeek-V4-Flash-0731. Many landed.

#56 opened about 1 month ago by

DeepSeel-v4-flash-0731 thinking aborted by EOS

#55 opened about 1 month ago by

Question for the DeepSeek team

#52 opened about 1 month ago by

vLLM + Ampere - DeepSeek V4 Flash 0731 working in a RTX3090 cluster

#51 opened about 1 month ago by

The depth-5 corruption reports were a serving-stack bug, not this model — resolution, guidance, and a small model-card suggestion

#50 opened about 1 month ago by

DeepSeek-V4-Flash-0731 在 4×A100 上部署：完整踩坑与工程复盘

#49 opened about 1 month ago by

PSA If you have terrible looping issues or random character generations try update your CUDA to 13.3

#46 opened about 1 month ago by

This model has many tool_call errors.

#45 opened about 1 month ago by

Here's how to run on SM120 / RTX 6000 Pro's (two of them) - not a troll/rickroll, I promise

#44 opened about 1 month ago by

Please add recipes and support for Blackwell (sm120) gpus

#43 opened about 1 month ago by

Update README.md

#42 opened about 2 months ago by

If the actual form of v4-Flash is already strong as such F**K, then what's final version of v4-Pro and WHEN it will be released?

#41 opened about 2 months ago by

# 0731 KV cache per-token is ~8× larger than preview (56 vs 7.2 bytes/token) - is this expected?

#40 opened about 2 months ago by

dfsafdsafxcvzxv

Reasoning loops

#39 opened about 2 months ago by

Add community evaluation results for DEEP-SWE

#38 opened about 2 months ago by

Tool-calling instruction is not consistently followed

#37 opened about 2 months ago by

[Ktransformers official] add Ktransformers cookbook to model card for consumer hardware

#36 opened about 2 months ago by

<3 to deepseek team....

#34 opened about 2 months ago by

VISION VISION VISION

#33 opened about 2 months ago by

Running 0731 on a single DGX Spark (GB10, 128GB) - working vLLM config, 256K context

#32 opened about 2 months ago by

Wen yah

#31 opened about 2 months ago by

Update README.md

#30 opened about 2 months ago by

Model returned random math problem when given nothing

#29 opened about 2 months ago by

[HF Staff] This tag is required to enable inference providers, please merge!

#28 opened about 2 months ago by

ds4

#27 opened about 2 months ago by

KakumuraArimura

Add missing chat_template.jinja

#26 opened about 2 months ago by

梁圣牛逼; deepseek is supercalifragilisticexpialidocious

#25 opened about 2 months ago by

震惊瘫坐，仿佛看到了原子弹爆炸！

#24 opened about 2 months ago by

GigaSeek

#23 opened about 2 months ago by

How can it be served with 4 x RTX Pro6000?

#22 opened about 2 months ago by

Preserve reasoning

#21 opened about 2 months ago by

Publish a blog or paper on how the improvements were achieved.

#18 opened about 2 months ago by

DGX Spark deploy

#17 opened about 2 months ago by

This should be named V4.1?

#16 opened about 2 months ago by

Are you saying I can deploy a top-tier T1 model with just a few thousand dollars' worth of Mac Studio?

#15 opened about 2 months ago by