by Black Forest Labs
State-of-the-art text-to-image model from Black Forest Labs. Excels at photorealism, text rendering, and prompt adherence. 12B parameter DiT architecture with dual text encoders: T5-XXL (4.7B) and CLIP-L (0.12B).
VRAM requirements, GPU fit, and setup notes for Flux.1 Dev, including 8GB/12GB fit guidance where relevant. Recommended runtimes: ComfyUI and Diffusers support. Best download size: ~6.8 GB at Q4_0.
Your hardware
Detecting...
Measured quality metrics for Flux.1 Dev outputs.
How often humans prefer this model's output (0-100%)
Visual quality and composition rating (5-9 scale)
Text-image alignment accuracy (higher is better)
Compare which GPUs can run Flux.1 Dev at different precisions. FP8 uses less memory than FP16 when available, and the grade shows how comfortably each GPU handles the workload.
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 23.6 GB | B● | F● | F● | F● |
| 768×768 | 23.8 GB | B● | F● | F● | F● |
| 1024×1024 | 24.0 GB | B● | F● | F● | F● |
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 12.6 GB | S● | B● | F● | S● |
| 768×768 | 12.8 GB | S● | B● | F● | S● |
| 1024×1024 | 13.0 GB | S● | B● | F● | S● |
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 12.6 GB | S● | B● | F● | S● |
| 768×768 | 12.8 GB | S● | B● | F● | S● |
| 1024×1024 | 13.0 GB | S● | B● | F● | S● |
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 13.8 GB | S | D | F | A |
| 768×768 | 14.1 GB | S | D | F | A |
| 1024×1024 | 14.5 GB | S | D | F | A |
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 6.7 GB | S● | S● | A● | S● |
| 768×768 | 6.8 GB | S● | S● | A● | S● |
| 1024×1024 | 7.0 GB | S● | S● | A● | S● |
GGUF Q4 available
Quantized GGUF format for lower VRAM and smaller downloads -- reduces download from 23.8 GB to 6.8 GB
Turbo / LCM distillation
Use distilled scheduler at 4-8 steps for faster iteration
ControlNets available
Add guided generation with 3 adapters (+3.6 GB VRAM each)
from diffusers import FluxPipeline
import torch
pipe = FluxPipeline.from_pretrained(
"black-forest-labs/FLUX.1-dev",
torch_dtype=torch.float16
)
pipe.to("cuda")
image = pipe(
prompt="your prompt here",
num_inference_steps=28,
guidance_scale=3.5,
height=1024,
width=1024,
).images[0]
image.save("output.png")Get started
Setup instructions for running Flux.1 Dev locally
1. Download the model
Get the checkpoint from HuggingFace
2. Place in:
ComfyUI/models/checkpoints/ (or ComfyUI/models/unet/ for GGUF)3. Launch ComfyUI
python main.pyComfyUI Workflow
Basic txt2img workflow for Flux.1 Dev
Drag & drop into ComfyUI or use File → Import
VRAM allocation at 1024×1024 on RTX 4090 24GB (24 GB)
Time per image at 1024×1024, 28 steps, FP16.
Download Flux.1 Dev in the precision that matches your GPU. Lower precision usually means less VRAM pressure, while higher precision keeps more quality.
| Format | Precision | Size | Provider | |
|---|---|---|---|---|
| Official Weights | ||||
| safetensorsRecommended | FP16 | 23.8 GB | official | Download |
| safetensors | FP8 | 11.9 GB | official | Download |
| Community Conversions | ||||
| ggufCommunity | Q2_K | 4.0 GB | community-gguf | Download |
| ggufCommunity | Q3_K_S | 5.2 GB | community-gguf | Download |
| ggufCommunity | Q4_0 | 6.8 GB | community-gguf | Download |
| ggufCommunity | Q4_K_S | 6.8 GB | community-gguf | Download |
| ggufCommunity | Q5_0 | 8.3 GB | community-gguf | Download |
| ggufCommunity | Q5_K_S | 8.3 GB | community-gguf | Download |
| ggufCommunity | Q6_K | 9.9 GB | community-gguf | Download |
| ggufCommunity | Q8_0 | 12.7 GB | community-gguf | Download |
3 ControlNets available for Flux.1 Dev. ControlNets add guided image generation (edges, depth, pose) at the cost of extra VRAM.
Extract edges from reference image to guide composition and structure. Best for architectural and product photography.
Use depth estimation to maintain 3D spatial relationships. Great for scenes with foreground/background separation.
Single model that handles canny, depth, pose, tile, and blur conditions. Most versatile option for Flux.
Growing ecosystem with hundreds of LoRAs on CivitAI and HuggingFace. Flux LoRAs are typically smaller than SDXL LoRAs due to the DiT architecture.
Approximately 500 LoRAs available on CivitAI. Each LoRA adds ~0.3 GB VRAM.
Frequently asked questions
Flux.1 Dev (12B parameters) requires approximately 24.0 GB of VRAM at FP16 precision for standard 1024×1024 image generation. If you want a lighter setup, lower precisions like FP8 can reduce memory use when available.
Yes, Flux.1 Dev can fit on some 8GB GPUs at ~6.8 GB at Q4_0. Check the VRAM table above for the exact resolution and precision trade-off.
Flux.1 Dev is marked for ComfyUI and Diffusers support in our catalog, so those are the runtimes we recommend first for local setup. If your workflow uses another front end, check the model's available formats and workflow notes above before downloading.
Flux.1 Dev can run on the RTX 4090 with sequential offloading enabled, though generation will be slower than native fit.
Yes, Flux.1 Dev has 3 ControlNet adapters available: Canny Edge, Depth Map, Union (Multi-Control). Each ControlNet adds roughly 3.6 GB of extra VRAM.
Growing ecosystem with hundreds of LoRAs on CivitAI and HuggingFace. Flux LoRAs are typically smaller than SDXL LoRAs due to the DiT architecture. The LoRA ecosystem for Flux.1 Dev is rated as "moderate". There are approximately 500 LoRAs available on Civitai. Each LoRA adds roughly 0.3 GB of extra VRAM.
On a reference GPU (RTX 4090 24GB), Flux.1 Dev generates a 1024×1024 image in approximately ~18s at FP16 with 28 inference steps. Faster GPUs with higher memory bandwidth will produce images more quickly.
See also