Skip to content

image model · flux · Windows

Can I run FLUX.1 dev on Nvidia GeForce GTX 1070 (8GB)?

Compatibility verdict VRAM check
Yes, but tight fast on this GPU

Yes. FLUX.1 dev runs on Nvidia GeForce GTX 1070 (8GB) at Q4 GGUF (~6.5 GB of ~7 GB usable).

Needs ~6.5 GB Device usable ~7 GB

Fits at Q4 GGUF (~6.5 GB of ~7 GB usable) but with little headroom. Close other apps to free a little VRAM before generating.

Peak VRAM
~6.5 GB
Usable on device
~7 GB
Device memory
8 GB
Quant
Q4 GGUF
Share on X Share on Reddit

How to run it

Use ComfyUI or Draw Things at Q4 GGUF. The big text encoder is loaded to encode your prompt, then offloaded before generation, which is why peak VRAM stays near the backbone size rather than the sum of every file.

Model flux
Type
image (DIT)
Parameters
12B
Peak VRAM
~6.5 GB at Q4 GGUF
Resolution
1024×1024
License
FLUX.1-dev Non-Commercial License
Full FLUX.1 dev requirements →
Device Windows
Memory
8 GB vram
Usable for weights
~7 GB
Power draw
~150 W
Best runtime
llama.cpp CUDA (Pascal, no tensor cores)
Best models for Nvidia GeForce GTX 1070 (8GB) →

You could also run

Run FLUX.1 dev on other hardware

FAQ

Can Nvidia GeForce GTX 1070 (8GB) run FLUX.1 dev?

Yes. FLUX.1 dev runs on Nvidia GeForce GTX 1070 (8GB) at Q4 GGUF (~6.5 GB of ~7 GB usable).

How much VRAM does FLUX.1 dev need?

It is a tight fit on Nvidia GeForce GTX 1070 (8GB). At Q4 GGUF the realistic peak is ~6.5 GB of VRAM, versus ~33 GB with every component kept resident (no offload). With aggressive CPU offload it drops to ~3 GB, much slower.

What do I use to run FLUX.1 dev locally?

FLUX.1 dev runs in ComfyUI or Draw Things (among others). It loads as a diffusion checkpoint plus its text encoder and VAE, not a single chat command.

Sources

VRAM figures are sourced peak-usage anchors at the noted quant, validated 2026-09-14. See methodology.