Model reference · open weights
FLUX.2-dev is an open-weight image model from city96. FLUX.2-dev-gguf (GGUF) weighs 20.1 GB; the smallest configuration that runs it is RTX 5090 32 GB.
Summary of the city96/FLUX.2-dev-gguf model card, 2026-10-01
What it is
| Released by | city96 |
|---|---|
| Released | 2025-11-25 |
| VRAM | 20.1 GB for the weights |
What it runs on
| Card | One image | Memory |
|---|---|---|
| RTX 3060 12 GB … RTX 4090 24 GB | does not fit | |
| RTX 5090 32 GB | fits | 31.0 GB |
| L40S 48 GB | fits | 44.0 GB |
| A100 80 GB | fits | 78.2 GB |
| H100 80 GB | fits | 78.1 GB |
| RTX PRO 6000 Blackwell 96 GB | fits | 93.8 GB |
| DGX Spark (GB10) 128 GB unified | fits | 107 GB |
| H200 141 GB | fits | 138 GB |
| B200 180 GB | fits | 176 GB |
From the model card
This is a direct GGUF conversion of black-forest-labs/FLUX.2-dev.
The model files can be used in ComfyUI with the ComfyUI-GGUF custom node. Place the required model(s) in the following folders:
| Type | Name | Location | Download |
|---|---|---|---|
| Main Model | flux2-dev | ComfyUI/models/diffusion_models | GGUF (this repo) |
| Text Encoder | Mistral-Small-3.2-24B-Instruct-2506 | ComfyUI/models/text_encoders | Safetensors / GGUF |
| VAE | flux2 VAE | ComfyUI/models/vae | Safetensors |
Example outputs - sample size of 1, not strictly representative
[!NOTE] As with Qwen-Image, Q5_K_M, Q4_K_M, Q3_K_M, Q3_K_S and Q2_K have some extra logic as to which blocks to keep in high precision.
The logic is partially based on guesswork, trial & error, and the graph found in the readme for Freepik/flux.1-lite-8B (which in turn quotes this blog by Ostris)
As this is a quantized model not a finetune, all the same restrictions/original license terms still apply.
Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.
How it works
Running it yourself
Rent a machine by the hour. ComfyUI is installed on it. Open ComfyUI through the tunnel, then Templates → Flux.2 Dev. Its model loader takes .safetensors files: replace it with Unet Loader (GGUF) and choose this model's file there.
# on your rented machine (the ssh line is on its page in the console)
# the GGUF loader nodes, once per machine
git clone https://github.com/city96/ComfyUI-GGUF /opt/ComfyUI/custom_nodes/ComfyUI-GGUF
pip install -r /opt/ComfyUI/custom_nodes/ComfyUI-GGUF/requirements.txt
# get REPO FILE FOLDER: one file into /workspace/models/FOLDER, where ComfyUI loads it from
get() { hf download "$1" "$2" --local-dir /workspace/hf-files && mkdir -p "/workspace/models/$3" && mv "/workspace/hf-files/$2" "/workspace/models/$3/$4"; }
# the model (18.7 GB)
get city96/FLUX.2-dev-gguf flux2-dev-Q4_K_M.gguf unet
# what the template also loads “Flux.2 Dev” (35.9 GB)
get black-forest-labs/FLUX.2-small-decoder full_encoder_small_decoder.safetensors vae
get Comfy-Org/flux2-dev split_files/text_encoders/mistral_3_small_flux2_bf16.safetensors text_encoders
get ByteZSzn/Flux.2-Turbo-ComfyUI Flux_2-Turbo-LoRA_comfyui.safetensors loras
start-comfyui# on your computer, in a second terminal: ComfyUI in your browser at http://localhost:8188
# HOST and PORT are your machine's, from its page in the console
ssh -L 8188:localhost:8188 dev@HOST -p PORT