briaai/Fibo-mlx-8bit

             

FIBO is the first open-source, JSON-native text-to-image model trained exclusively on long structred captions.

Fibo sets a new standard for controllability, predictability, and disentanglement.

Fibo MLX 8-bit

This is an 8-bit quantization of FIBO made with mflux.

Run SOTA on Your Mac

This quantization brings the state-of-the-art capabilities of FIBO to your local machine. Thanks to the 8-bit quantization and FIBO's efficient architecture, you can now run this powerful 8-billion-parameter model directly on your Mac with excellent performance and reduced memory usage.

About the Project

FIBO

FIBO is an 8B parameter DiT-based, flow-matching text-to-image model designed for enterprise-ready applications. It focuses on:

• Controllability & Predictability: Exceptional prompt adherence using long, structured JSON captions.
• Responsible AI: Trained exclusively on licensed data.
• Architecture: Features SmolLM3-3B as the text encoder and Wan 2.2 as the VAE.

Mflux

Mflux is a powerful generation framework built on top of Apple's MLX. It enables efficient inference of diffusion models on Apple Silicon, unlocking the full potential of your Mac's hardware.

License & Access

The weights for the original FIBO model are open source for non-commercial use. Please refer to the original model card for access and license details.