Instructions to use Zeraix/llama-seeds with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Zeraix/llama-seeds 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 Zeraix/llama-seeds:Q4_1
# Run inference directly in the terminal:
llama cli -hf Zeraix/llama-seeds:Q4_1

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Zeraix/llama-seeds:Q4_1
# Run inference directly in the terminal:
llama cli -hf Zeraix/llama-seeds:Q4_1

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 Zeraix/llama-seeds:Q4_1
# Run inference directly in the terminal:
./llama-cli -hf Zeraix/llama-seeds:Q4_1

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 Zeraix/llama-seeds:Q4_1
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Zeraix/llama-seeds:Q4_1

Use Docker

docker model run hf.co/Zeraix/llama-seeds:Q4_1

LM Studio
Jan
Ollama
How to use Zeraix/llama-seeds with Ollama:
```
ollama run hf.co/Zeraix/llama-seeds:Q4_1
```
Unsloth Desktop

How to use Zeraix/llama-seeds with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Zeraix/llama-seeds:Q4_1

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": "Zeraix/llama-seeds:Q4_1"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Zeraix/llama-seeds with Docker Model Runner:
```
docker model run hf.co/Zeraix/llama-seeds:Q4_1
```

Lemonade

How to use Zeraix/llama-seeds with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Zeraix/llama-seeds:Q4_1

Run and chat with the model

lemonade run user.llama-seeds-Q4_1

List all available models

lemonade list

Hermes Agent

How to use Zeraix/llama-seeds with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Zeraix/llama-seeds:Q4_1

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 Zeraix/llama-seeds:Q4_1

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Zeraix/llama-seeds with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Zeraix/llama-seeds:Q4_1

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 "Zeraix/llama-seeds:Q4_1" \
  --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"

Zeraix prefix-cache seeds and auxiliary assets

This repository holds assets consumed by the Zeraix desktop application. It is not a single standalone chat model: cache archives and auxiliary drafter files have different roles and compatibility requirements.

What is here?

Prefix-cache seeds: precomputed KV state for the application's initial prompt and tool declarations. A matching seed can avoid recomputing that prefix on the first request.
Auxiliary drafters: files under the drafters/ directory for specific speculative-decoding integrations. A drafter is not a replacement for its target model.

Compatibility matters

A seed must match the prompt/tool-prefix hash, pinned model revision, KV disk-format version, KV quantization, and the runtime's model/cache identity. The application selects compatible assets; do not mix archives across model or runtime versions.

The currently documented prefix-seed integration is macOS-specific. Its presence here does not imply equivalent Windows support. If a seed is unavailable or cannot be installed, the application is designed to fall back to a cold prefill.

See the public seed installer and identity rules and model configuration for exact version selection and drafter pairing.

Use through Zeraix

Download Zeraix and let the application manage these downloads. Generic Hub-generated model-loading commands do not describe how to load this mixed asset repository correctly.

Existing repository names, revisions, and paths are retained for application compatibility. This documentation update does not replace any cache archive or model file.

Related projects

Runtime and sandbox assets
Zeraix source and documentation
Imparo inference engine, a separate project with its own state system and validation scope.

Licenses and provenance

Model-derived assets and auxiliary weights retain the applicable upstream model and component terms. This repository-level description does not relicense those files. Consult the model configuration, pinned source revisions, and relevant upstream model cards before redistributing an artifact.

For compatibility or provenance questions, open a Zeraix issue with the exact asset path and application version.

Downloads last month: 34

GGUF

Model size

1B params

Architecture

dflash

Hardware compatibility

4-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support