BajanYSL/FLUX.1-dev

1---2title: FLUX.1 dev3emoji: 🖼️4colorFrom: yellow5colorTo: purple6sdk: static7app_file: index.html8short_description: Text-to-image with FLUX.1 [dev] via Inference Providers9pinned: false10hf_oauth: true11hf_oauth_expiration_minutes: 48012hf_oauth_scopes:13  - inference-api14  - gated-repos15---16 17# FLUX.1 [dev]18 19Text-to-image demo of [black-forest-labs/FLUX.1-dev](https://huggingface.co/black-forest-labs/FLUX.1-dev),20a 12B rectified-flow transformer guidance-distilled from FLUX.1 [pro].21 22## How it works23 24This is a **static** Space — there is no server and no GPU. The page calls25[Inference Providers](https://huggingface.co/docs/inference-providers) through the26Hugging Face router, which routes to fal.ai, Replicate or WaveSpeed.27 28Sign in with your Hugging Face account and generation is billed to **your** own29included inference credits, not the Space owner's. Nothing is proxied through a30shared key, and your token stays in your browser.31 32FLUX.1 [dev] is released under a33[non-commercial license](https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md).34 35## Why static and not ZeroGPU36 37Self-hosting a 12B model needs a GPU. ZeroGPU requires a PRO subscription or an38account older than 30 days, so this demo routes to Inference Providers instead.39The self-hosted Gradio + ZeroGPU build is parked in [`zerogpu/`](./zerogpu) with40switch-over instructions — none of it runs today.41