Instructions to use Guilherme34/Firefly-v4-BetterGGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Guilherme34/Firefly-v4-BetterGGUF with Transformers:

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("Guilherme34/Firefly-v4-BetterGGUF", dtype="auto")

llama-cpp-python

How to use Guilherme34/Firefly-v4-BetterGGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="Guilherme34/Firefly-v4-BetterGGUF",
	filename="Firefly-v4.IQ4_XS.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use Guilherme34/Firefly-v4-BetterGGUF 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 Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Guilherme34/Firefly-v4-BetterGGUF: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 Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf Guilherme34/Firefly-v4-BetterGGUF: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 Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M

Use Docker

docker model run hf.co/Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M

LM Studio
Jan
Ollama
How to use Guilherme34/Firefly-v4-BetterGGUF with Ollama:
```
ollama run hf.co/Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M
```

Unsloth Studio

How to use Guilherme34/Firefly-v4-BetterGGUF 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 Guilherme34/Firefly-v4-BetterGGUF 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 Guilherme34/Firefly-v4-BetterGGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Guilherme34/Firefly-v4-BetterGGUF to start chatting

How to use Guilherme34/Firefly-v4-BetterGGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Guilherme34/Firefly-v4-BetterGGUF: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": "Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use Guilherme34/Firefly-v4-BetterGGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Guilherme34/Firefly-v4-BetterGGUF: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 Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use Guilherme34/Firefly-v4-BetterGGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Guilherme34/Firefly-v4-BetterGGUF: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 "Guilherme34/Firefly-v4-BetterGGUF: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 Guilherme34/Firefly-v4-BetterGGUF with Docker Model Runner:
```
docker model run hf.co/Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M
```

Lemonade

How to use Guilherme34/Firefly-v4-BetterGGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Guilherme34/Firefly-v4-BetterGGUF:Q4_K_M

Run and chat with the model

lemonade run user.Firefly-v4-BetterGGUF-Q4_K_M

List all available models

lemonade list

Firefly-v4 GGUF

Polished GGUF quantizations of Firefly-v4, with experimental Multi-Token Prediction support for compatible inference backends.

GGUF MTP Vision Audio Gemma 4 Roleplay Reasoning

Format

GGUF

Feature

MTP Ready

Focus

Creative

Mode

Multimodal

About

This repository contains GGUF conversions of Firefly-v4, a multimodal fine-tune based on p-e-w/gemma-4-E2B-it-heretic-ara.

These files are intended for local inference with GGUF-compatible software such as:

llama.cpp LM Studio KoboldCpp Jan Text Generation WebUI Ollama

Original model: https://huggingface.co/Guilherme34/Firefly-v4

MTP Support

Includes experimental Multi-Token Prediction support. Performance and compatibility depend on your inference backend.

Multimodal

Retains Gemma 4's native multimodal capabilities, including image and audio understanding.

Creative Focus

Optimized for creative writing, roleplay, long conversations, and general instruction following.

Optional Reasoning

Supports the <|think|> tag for switchable reasoning on compatible backends.

A Small Note

This GGUF release includes experimental MTP support.

I was not originally planning to spend much time on Firefly-v4's GGUF release, but many people have enjoyed using the model. Because of that, I decided to give this repository the attention it deserves and polish it properly.

That said, I still personally consider Firefly-v3 to be my best model overall. Everyone has different preferences, though, and I am happy to continue supporting Firefly-v4 for those who enjoy it.

Thank you to everyone who has shared feedback and supported the project.

Quant	Recommendation
Q4_K_M	Recommended for most users
Q4_K_S	Fast and efficient
Q6_K	Excellent quality
Q8_0	Highest GGUF quality
F16	Maximum precision

Available

Type	Notes
Q2_K	Smallest size
Q3_K_S	Low-memory systems
Q3_K_M	Balanced
Q3_K_L	Improved Q3 quality
IQ4_XS	Excellent efficiency
Q4_K_S	Fast
Q4_K_M	Recommended
Q5_K_S	Higher quality
Q5_K_M	Higher quality
Q6_K	Excellent quality
Q8_0	Near-lossless
F16	Full precision

Attribution

Original Model: Firefly-v4
Author: Guilherme34

Base Model: p-e-w/gemma-4-E2B-it-heretic-ara

This repository contains only GGUF conversions. All credit for the model training, fine-tuning, datasets, and original work belongs to the original author.

License

This repository is distributed under the Firefly-v4 Attribution License.

Commercial use, quantizations, merges, and derivative works are permitted under the terms described in the included LICENSE file.

Please preserve the required attribution when redistributing or building upon this model.

⚠️ Disclaimer

This model is uncensored and will generate content without built-in refusals. It is intended for creative fiction and roleplay between consenting adults. The creator is not responsible for how the model is used. Do not use it to produce content that is illegal in your jurisdiction.

Thank you for using Firefly-v4 GGUF

If you enjoy this release, consider checking out the original model and supporting future development.

Original Model: https://huggingface.co/Guilherme34/Firefly-v4

Downloads last month: 1,625

GGUF

Model size

5B params

Architecture

gemma4

Hardware compatibility

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

16-bit

View +1 variant

Inference Providers NEW

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

Model tree for Guilherme34/Firefly-v4-BetterGGUF

Base model

p-e-w/gemma-4-E2B-it-heretic-ara

Finetuned

Guilherme34/Firefly-v4

Quantized

(4)

this model

Firefly-v4 GGUF

About

MTP Support

Multimodal

Creative Focus

Optional Reasoning

A Small Note

Recommended

Available

Attribution

License

Thank you for using Firefly-v4 GGUF

Model tree for Guilherme34/Firefly-v4-BetterGGUF