Fibo 2.0 GGUF

This repository contains GGUF quantizations of Fibo 2.0 Dev and Fibo 2.0 Edit Turbo. The Dev variant supports text-to-image and image editing. Edit Turbo is distilled for four-step image editing and is not intended for text-to-image.

Both variants use the Fibo 2.0 architecture. Dev uses 50 steps and CFG 5 for text-to-image. Edit Turbo uses 4 steps and CFG 1 for image editing.

Example outputs

Fibo 2.0 Dev

Fibo 2.0 Dev text-to-image example

Fibo 2.0 Edit Turbo

Fibo 2.0 Edit Turbo image-to-image example

The edit workflow uses the included fibo2-test-input.png. Copy it to ComfyUI/input/, or replace it with your own image in the workflow's Load Image node.

Quantizations

File sizes are shown in GiB. Standard GGUF files use common GGML formats. Q8_CR uses the custom ConvRot layout and requires comfyui-gguf-reboot.

Variant Quant Format Size File
Dev Q8_0 Standard GGUF 8.29 GiB fibo2-dev-Q8_0.gguf
Dev Q6_K Standard GGUF 6.48 GiB fibo2-dev-Q6_K.gguf
Dev Q4_K_M Standard GGUF 5.16 GiB fibo2-dev-Q4_K_M.gguf
Dev Q8_CR Proprietary, comfyui-gguf-reboot 7.83 GiB fibo2-dev_Q8_CR.gguf
Edit Turbo Q8_0 Standard GGUF 8.29 GiB fibo2-edit-turbo-Q8_0.gguf
Edit Turbo Q6_K Standard GGUF 6.48 GiB fibo2-edit-turbo-Q6_K.gguf
Edit Turbo Q4_K_M Standard GGUF 5.16 GiB fibo2-edit-turbo-Q4_K_M.gguf
Edit Turbo Q8_CR Proprietary, comfyui-gguf-reboot 7.83 GiB fibo2-edit-turbo_Q8_CR.gguf

ComfyUI setup

Fibo2 requires ComfyUI core commit 87ed5a066d9c4660ef42100fd01a888da6994856 or newer, plus comfyui-gguf-reboot. Load the GGUF with UnetLoaderGGUFDynamicVRAM from models/diffusion_models/.

Download the Qwen3-VL 4B Int8 ConvRot text encoder and place it at ComfyUI/models/text_encoders/qwen3vl_4b_int8_convrot.safetensors.

Download the FLUX.2 VAE and save it as ComfyUI/models/vae/fibo2-flux2-vae.safetensors so it does not replace another FLUX.2 VAE.

Workflows

License and attribution

Fibo 2.0 Dev and Edit Turbo are released under CC BY-NC 4.0. Commercial use requires a separate agreement with Bria. See the Dev model card and Edit Turbo model card for the complete terms and model details.

Downloads last month
908
GGUF
Model size
8B params
Architecture
fibo2
Hardware compatibility
Log In to add your hardware

4-bit

6-bit

8-bit

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

Model tree for molbal/fibo-2.0-GGUF

Quantized
(14)
this model