Instructions to use Nock-AI/nock-agent-31b-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Nock-AI/nock-agent-31b-GGUF 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 Nock-AI/nock-agent-31b-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Nock-AI/nock-agent-31b-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Nock-AI/nock-agent-31b-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Nock-AI/nock-agent-31b-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf Nock-AI/nock-agent-31b-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf Nock-AI/nock-agent-31b-GGUF:Q4_K_M

Use Docker

docker model run hf.co/Nock-AI/nock-agent-31b-GGUF:Q4_K_M

Llama new
LM Studio
Jan
Ollama
How to use Nock-AI/nock-agent-31b-GGUF with Ollama:
```
ollama run hf.co/Nock-AI/nock-agent-31b-GGUF:Q4_K_M
```
Unsloth Desktop

How to use Nock-AI/nock-agent-31b-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Nock-AI/nock-agent-31b-GGUF:Q4_K_M

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": "Nock-AI/nock-agent-31b-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Nock-AI/nock-agent-31b-GGUF with Docker Model Runner:
```
docker model run hf.co/Nock-AI/nock-agent-31b-GGUF:Q4_K_M
```

Lemonade

How to use Nock-AI/nock-agent-31b-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Nock-AI/nock-agent-31b-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.nock-agent-31b-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use Nock-AI/nock-agent-31b-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Nock-AI/nock-agent-31b-GGUF:Q4_K_M

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 Nock-AI/nock-agent-31b-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Nock-AI/nock-agent-31b-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Nock-AI/nock-agent-31b-GGUF:Q4_K_M

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 "Nock-AI/nock-agent-31b-GGUF:Q4_K_M" \
  --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"

Nock Agent 31B (GGUF)

Nock Agent 31B as a single GGUF file for Ollama and llama.cpp: the LoRA from Nock-AI/nock-agent-31b-lora merged into Gemma 4 31B and quantized to Q4_K_M.

File	Quant	Size
`nock-agent-31b-Q4_K_M.gguf`	Q4_K_M	about 19 GB

Plan for roughly 24 GB of free memory (GPU, unified memory, or system RAM; CPU only works but is slow).

Run it

With llama.cpp (tested):

llama-server -hf Nock-AI/nock-agent-31b-GGUF:Q4_K_M --jinja -ngl 99 -c 8192

Then call the OpenAI compatible endpoint (/v1/chat/completions) with your tools. --jinja uses the model's own chat template, which is needed for tool calls.

With Ollama (needs an Ollama version that supports Gemma 4):

ollama run hf.co/Nock-AI/nock-agent-31b-GGUF:Q4_K_M

The system prompt it was trained with:

You are Nock Agent, a coding agent working in a project folder. Use the tools to read the code, make the smallest correct edit, and run the tests. Never edit test files. When the tests pass, reply with a one line summary and stop calling tools.

The model thinks before it answers (Gemma 4 reasoning). Give it room: with a small max_tokens the answer can come back empty because the budget went to thinking. 1,000 or more is a safe start for plain questions.

Checked

This file was loaded with llama.cpp (llama-server --jinja, all layers on one A100 80 GB, about 44 tokens per second). Given our system prompt, two tools and a bug report, its first action was the right one: a run_command call to run the tests before touching any code. A plain coding question got a correct answer.

The results on the LoRA model card were measured with the full model served in FP8 with vLLM (145 of 150 agent runs solved, 0 test files edited). The 150-run agent test was not repeated on this Q4_K_M file, and 4-bit quantization can change results.

Limits

Measured on small Python projects (2 to 6 files). Large repositories were not tested.
It can miss a fix that has to be repeated across many files. Search the project when a rename or a moved function is involved.
Not a security tool. Do not use it to audit code.