Instructions to use Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS
# Run inference directly in the terminal:
llama cli -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS
# Run inference directly in the terminal:
llama cli -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS
# Run inference directly in the terminal:
./llama-cli -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Use Docker

docker model run hf.co/Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

LM Studio
Jan

vLLM

How to use Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Ollama
How to use Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS with Ollama:
```
ollama run hf.co/Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS
```
Unsloth Desktop

How to use Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS with Docker Model Runner:
```
docker model run hf.co/Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS
```

Lemonade

How to use Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Run and chat with the model

lemonade run user.DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS-{{QUANT_TAG}}

List all available models

lemonade list

Hermes Agent

How to use Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

DeepSeek-V4-Flash-0731 – Custom IQ3_XXS / IQ2_XXS GGUF

This repository contains a custom mixed-quantization GGUF of DeepSeek-V4-Flash-0731.

The model was created from the original DeepSeek-V4-Flash-0731 GGUF and uses an experimental mixed quantization recipe designed to significantly reduce memory usage and increase inference speed on NVIDIA Blackwell GPUs.

Experimental quantization: This model has not been extensively evaluated for quality. Please benchmark it against the original model before using it for production workloads.

Base Model

DeepSeek-V4-Flash-0731

Original model: https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731

License: MIT

This repository contains a derived/quantized GGUF representation of the original model.

Quantization Recipe

The model uses the following custom tensor-type recipe:

^token_embd\.weight$=Q8_0
^output\.weight$=Q8_0
^blk\.[0-9]+\.(attn|indexer).*\.weight$=Q6_K
^blk\.[0-9]+\.ffn_(gate|up|down)_shexp\.weight$=Q6_K
^.*hc_.*\.weight$=Q8_0
^blk\.(0|1|2|3|4|5|6|7|8|9|34|35|36|37|38|39|40|41|42)\.ffn_(gate|up|down)_exps\.weight$=IQ3_XXS
^blk\.(10|11|12|13|14|15|16|17|18|19|20|21|22|23|24|25|26|27|28|29|30|31|32|33)\.ffn_(gate|up|down)_exps\.weight$=IQ2_XXS

Downloads last month: 38

GGUF

Model size

284B params

Architecture

deepseek4

Hardware compatibility

We're not able to determine the quantization variants.

View all variants

Model tree for Oxidez/DeepSeek-V4-Flash-0731-UD-IQ3_IQ2-XS

Base model

deepseek-ai/DeepSeek-V4-Flash-0731

Quantized

(193)

this model