Instructions to use SwarmDo/SwarmDo-A1-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SwarmDo/SwarmDo-A1-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="SwarmDo/SwarmDo-A1-GGUF",
	filename="SwarmDo-A1-Q4_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use SwarmDo/SwarmDo-A1-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 SwarmDo/SwarmDo-A1-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf SwarmDo/SwarmDo-A1-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 SwarmDo/SwarmDo-A1-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf SwarmDo/SwarmDo-A1-GGUF:Q4_K_M

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 SwarmDo/SwarmDo-A1-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf SwarmDo/SwarmDo-A1-GGUF:Q4_K_M

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 SwarmDo/SwarmDo-A1-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf SwarmDo/SwarmDo-A1-GGUF:Q4_K_M

Use Docker

docker model run hf.co/SwarmDo/SwarmDo-A1-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use SwarmDo/SwarmDo-A1-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SwarmDo/SwarmDo-A1-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SwarmDo/SwarmDo-A1-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SwarmDo/SwarmDo-A1-GGUF:Q4_K_M

Ollama
How to use SwarmDo/SwarmDo-A1-GGUF with Ollama:
```
ollama run hf.co/SwarmDo/SwarmDo-A1-GGUF:Q4_K_M
```

Unsloth Studio

How to use SwarmDo/SwarmDo-A1-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for SwarmDo/SwarmDo-A1-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for SwarmDo/SwarmDo-A1-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for SwarmDo/SwarmDo-A1-GGUF to start chatting

How to use SwarmDo/SwarmDo-A1-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf SwarmDo/SwarmDo-A1-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "SwarmDo/SwarmDo-A1-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use SwarmDo/SwarmDo-A1-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 SwarmDo/SwarmDo-A1-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 SwarmDo/SwarmDo-A1-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use SwarmDo/SwarmDo-A1-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf SwarmDo/SwarmDo-A1-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 "SwarmDo/SwarmDo-A1-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"

Docker Model Runner
How to use SwarmDo/SwarmDo-A1-GGUF with Docker Model Runner:
```
docker model run hf.co/SwarmDo/SwarmDo-A1-GGUF:Q4_K_M
```

Lemonade

How to use SwarmDo/SwarmDo-A1-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull SwarmDo/SwarmDo-A1-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.SwarmDo-A1-GGUF-Q4_K_M

List all available models

lemonade list

SwarmDo-A1 — GGUF (Ollama / llama.cpp / LM Studio)

Run a strong open coding model locally in one command. SwarmDo-A1 is an execution-grounded coding model built on Qwen3.6-27B (Apache-2.0). These are ready-to-run GGUF quantizations for Ollama, llama.cpp, and LM Studio — no Python, no setup.

ollama run hf.co/SwarmDo/SwarmDo-A1-GGUF:Q4_K_M

That's it — Ollama pulls the model straight from Hugging Face. No ollama.com account needed.

Quick start

Ollama (recommended):

ollama run hf.co/SwarmDo/SwarmDo-A1-GGUF:Q4_K_M     # ~16 GB, best size/quality balance
ollama run hf.co/SwarmDo/SwarmDo-A1-GGUF:Q8_0       # ~29 GB, higher fidelity

llama.cpp:

llama-cli -m SwarmDo-A1-Q4_K_M.gguf -p "Write a Python LRU cache with a capacity limit." -n 512

LM Studio: search SwarmDo/SwarmDo-A1-GGUF in the in-app model catalog and download a quant.

Which quant should I pick?

Tag	File	Size	Min RAM/VRAM*	Use it when
`Q4_K_M`	`SwarmDo-A1-Q4_K_M.gguf`	~16.5 GB	~18–20 GB	Default. Best balance of size, speed, quality.
`Q8_0`	`SwarmDo-A1-Q8_0.gguf`	~28.6 GB	~30–32 GB	Maximum fidelity, near-lossless vs fp16.

*Approximate; leave headroom for context. Q4_K_M runs on a single 24 GB GPU or a 32 GB Mac.

What is SwarmDo-A1?

SwarmDo-A1 is a coding model whose edge comes from execution-grounded development — it is tuned and evaluated against real test execution, not just leaderboard text matching. On agentic software-engineering tasks, our re-based + tuned stack shows a decision-grade paired win over its starting point (~+25 points, McNemar p < 0.01, measured in-harness). It is built on the Qwen3.6-27B linear-attention hybrid base (Apache-2.0), so it's fully open and self-hostable.

Strengths: Python and general-purpose code generation, bug-fixing, and agentic coding workflows.
License: Apache-2.0 — commercial use allowed.
Context: long-context capable (linear-attention hybrid architecture).

Honest scope (please read)

A bare GGUF gives you the model's core text/code generation. The full SwarmDo-A1 system — the execution-verified best-of-N selection and the render-verified visual-coding capability (image → code) — runs through the complete self-hosted harness (vLLM + tools + vision tower), which a plain chat GGUF does not carry. For the full system see SwarmDo/SwarmDo-A1. If you just want a fast, strong local coding chat model, this GGUF is exactly that.

Citation

@misc{swarmdo-a1,
  title  = {SwarmDo-A1: an execution-grounded open coding model},
  author = {SwarmDo},
  year   = {2026},
  url    = {https://huggingface.co/SwarmDo/SwarmDo-A1}
}

Downloads last month: 304

GGUF

Model size

27B params

Architecture

qwen35

Hardware compatibility

4-bit

8-bit

Model tree for SwarmDo/SwarmDo-A1-GGUF

Base model

Qwen/Qwen3.6-27B

Quantized

(642)

this model

Evaluation results

paired win vs base (McNemar) on SWE-bench (agentic, in-harness)
self-reported

+25pp re-base, p<0.01

SwarmDo
/

SwarmDo-A1-GGUF

SwarmDo-A1 — GGUF (Ollama / llama.cpp / LM Studio)

Quick start

Which quant should I pick?

What is SwarmDo-A1?

Honest scope (please read)

Links

Citation

Model tree for SwarmDo/SwarmDo-A1-GGUF

Evaluation results