Trinity-Large-Thinking-GGUF

Instructions to use arcee-ai/Trinity-Large-Thinking-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use arcee-ai/Trinity-Large-Thinking-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="arcee-ai/Trinity-Large-Thinking-GGUF",
	filename="Trinity-Large-Thinking-IQ1_M/Trinity-Large-Thinking-IQ1_M-00001-of-00003.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use arcee-ai/Trinity-Large-Thinking-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama-cli -hf arcee-ai/Trinity-Large-Thinking-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 arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf arcee-ai/Trinity-Large-Thinking-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 arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M

Use Docker

docker model run hf.co/arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use arcee-ai/Trinity-Large-Thinking-GGUF with Ollama:
```
ollama run hf.co/arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M
```

Unsloth Studio new

How to use arcee-ai/Trinity-Large-Thinking-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 arcee-ai/Trinity-Large-Thinking-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 arcee-ai/Trinity-Large-Thinking-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for arcee-ai/Trinity-Large-Thinking-GGUF to start chatting

Pi new

How to use arcee-ai/Trinity-Large-Thinking-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf arcee-ai/Trinity-Large-Thinking-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": "arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use arcee-ai/Trinity-Large-Thinking-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf arcee-ai/Trinity-Large-Thinking-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 arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M

Run Hermes

hermes

Docker Model Runner
How to use arcee-ai/Trinity-Large-Thinking-GGUF with Docker Model Runner:
```
docker model run hf.co/arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M
```

Lemonade

How to use arcee-ai/Trinity-Large-Thinking-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Trinity-Large-Thinking-GGUF-Q4_K_M

List all available models

lemonade list

How to use from the

Use from the

llama-cpp-python library

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="arcee-ai/Trinity-Large-Thinking-GGUF",
	filename="",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Trinity-Large-Thinking-GGUF

Introduction

Trinity-Large-Thinking is a reasoning-optimized variant of Arcee AI's Trinity-Large family — a 398B-parameter sparse Mixture-of-Experts (MoE) model with approximately 13B active parameters per token, post-trained with extended chain-of-thought reasoning and agentic RL.

This repository contains GGUF quantized weights of Trinity-Large-Thinking in multiple quantization levels.

For full model details, benchmarks, and usage guidance, see the main Trinity-Large-Thinking model card.

Available Quantizations

Quant	Type	Use Case
Q8_0	8-bit	Best quality, highest memory
Q6_K_L	6-bit (large)	Near-lossless
Q6_K	6-bit	Near-lossless
Q5_K_L	5-bit (large)	High quality
Q5_K_M	5-bit (medium)	High quality
Q5_K_S	5-bit (small)	High quality
Q4_K_L	4-bit (large)	Recommended balance of quality and size
Q4_K_M	4-bit (medium)	Good balance
Q4_K_S	4-bit (small)	Good balance
Q4_1	4-bit	Good balance
Q4_0	4-bit	Good balance
Q3_K_XL	3-bit (extra large)	Lower memory
Q3_K_L	3-bit (large)	Lower memory
Q3_K_M	3-bit (medium)	Lower memory
Q3_K_S	3-bit (small)	Lower memory
IQ4_NL	4-bit (imatrix)	Importance-weighted 4-bit
IQ4_XS	4-bit (imatrix)	Importance-weighted 4-bit, smaller
IQ3_M	3-bit (imatrix)	Importance-weighted 3-bit
IQ3_XS	3-bit (imatrix)	Importance-weighted 3-bit, smaller
IQ3_XXS	3-bit (imatrix)	Importance-weighted 3-bit, smallest
IQ2_M	2-bit (imatrix)	Extreme compression
IQ2_S	2-bit (imatrix)	Extreme compression
IQ2_XS	2-bit (imatrix)	Extreme compression
IQ2_XXS	2-bit (imatrix)	Extreme compression
Q2_K_L	2-bit (large)	Extreme compression
Q2_K	2-bit	Extreme compression
IQ1_M	1-bit (imatrix)	Research / experimental
IQ1_S	1-bit (imatrix)	Research / experimental

Usage

llama.cpp

Supported in llama.cpp release b7061+.

# Recommended quant
llama-server -hf arcee-ai/Trinity-Large-Thinking-GGUF:Q4_K_M

# Higher quality
llama-server -hf arcee-ai/Trinity-Large-Thinking-GGUF:Q6_K

# Lower memory
llama-server -hf arcee-ai/Trinity-Large-Thinking-GGUF:Q3_K_M

LM Studio

Search for arcee-ai/Trinity-Large-Thinking-GGUF in Model Search. Select your preferred quantization level.

API

Works out of the box on OpenRouter as arcee-ai/trinity-large-thinking.

License

Trinity-Large-Thinking-GGUF is released under the Apache License, Version 2.0.

Citation

If you use this model, please cite:

@misc{singh2026arceetrinity,
  title        = {Arcee Trinity Large Technical Report},
  author       = {Varun Singh and Lucas Krauss and Sami Jaghouar and Matej Sirovatka and Charles Goddard and Fares Obied and Jack Min Ong and Jannik Straube and Fern and Aria Harley and Conner Stewart and Colin Kealty and Maziyar Panahi and Simon Kirsten and Anushka Deshpande and Anneketh Vij and Arthur Bresnu and Pranav Veldurthi and Raghav Ravishankar and Hardik Bishnoi and DatologyAI Team and Arcee AI Team and Prime Intellect Team and Mark McQuade and Johannes Hagemann and Lucas Atkins},
  year         = {2026},
  eprint       = {2602.17004},
  archivePrefix= {arXiv},
  primaryClass = {cs.LG},
  doi          = {10.48550/arXiv.2602.17004},
  url          = {https://arxiv.org/abs/2602.17004}
}