Instructions to use laion/marin-32b-base-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use laion/marin-32b-base-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="laion/marin-32b-base-GGUF",
	filename="marin-32b-base-Q4_K_M.gguf",
)

output = llm(
	"Once upon a time,",
	max_tokens=512,
	echo=True
)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use laion/marin-32b-base-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 laion/marin-32b-base-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf laion/marin-32b-base-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 laion/marin-32b-base-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf laion/marin-32b-base-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 laion/marin-32b-base-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf laion/marin-32b-base-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 laion/marin-32b-base-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf laion/marin-32b-base-GGUF:Q4_K_M

Use Docker

docker model run hf.co/laion/marin-32b-base-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use laion/marin-32b-base-GGUF with Ollama:
```
ollama run hf.co/laion/marin-32b-base-GGUF:Q4_K_M
```

Unsloth Studio

How to use laion/marin-32b-base-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 laion/marin-32b-base-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 laion/marin-32b-base-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for laion/marin-32b-base-GGUF to start chatting

Atomic Chat new
Docker Model Runner
How to use laion/marin-32b-base-GGUF with Docker Model Runner:
```
docker model run hf.co/laion/marin-32b-base-GGUF:Q4_K_M
```

Lemonade

How to use laion/marin-32b-base-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull laion/marin-32b-base-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.marin-32b-base-GGUF-Q4_K_M

List all available models

lemonade list

marin-32b-base-GGUF

GGUF quantizations of marin-community/marin-32b-base, a 32B dense base model (Qwen3 architecture).

These were produced with llama.cpp (convert_hf_to_gguf.py → f16 GGUF → llama-quantize).

Available quantizations

File	Quant	Notes
`marin-32b-base-Q4_K_M.gguf`	Q4_K_M	4-bit, medium — good size/quality balance (recommended default)
`marin-32b-base-Q5_K_M.gguf`	Q5_K_M	5-bit, medium — higher quality, larger
`marin-32b-base-Q6_K.gguf`	Q6_K	6-bit — near-lossless
`marin-32b-base-Q8_0.gguf`	Q8_0	8-bit — effectively lossless vs f16

Source

Base model: marin-community/marin-32b-base (Apache-2.0)
Architecture: Qwen3 (Qwen3ForCausalLM), 64 layers, hidden 5120, vocab 128256
This is a base (non-instruct) model; there is no chat template.

Usage (llama.cpp)

./llama-cli -m marin-32b-base-Q4_K_M.gguf -p "Your prompt here"

Downloads last month: 248

GGUF

Model size

33B params

Architecture

qwen3

Hardware compatibility

4-bit

5-bit

6-bit

8-bit

Inference Providers NEW

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

Model tree for laion/marin-32b-base-GGUF

Base model

marin-community/marin-32b-base

Quantized

(6)

this model