Flux
Black Forest Labs' open-weight image model — photorealistic, runs locally or via API.
Add Flux to your hut →Flux is Black Forest Labs' family of open-weight image generation models, built by the team that created Stable Diffusion. It produces highly photorealistic images with strong detail, accurate anatomy, and natural lighting — areas where earlier diffusion models struggled. The weights are publicly released, so Flux can be run locally on a capable GPU or accessed through inference providers like Replicate, fal.ai, and Together AI. Flux.1 [dev] and [schnell] are the most widely used variants; [pro] is the commercial API-only version.
Cost for self-hosting is just compute. API access through third-party providers is usage-based, typically fractions of a cent per image. Most often compared to Stable Diffusion (older, more ecosystem tooling) and Midjourney (closed, higher aesthetic quality out of the box but no local option).
| Made by | Black Forest Labs |
|---|---|
| Pricing | Open-weight (free to self-host) · API via providers (usage-based, ~$0.003–0.05/image) |
| Best for | Photorealistic images, local generation, developer API integration |
Alternatives to Flux
- Midjourney
Best-in-class stylised image generation via Discord or web — photorealism and art styles.
- DALL-E
OpenAI's image model — integrated into ChatGPT, strong at instruction-following.
- Ideogram
Image generator with best-in-class text rendering inside images — great for design assets.
- Civitai
Community hub for open-source image models — browse, download, and run Stable Diffusion checkpoints and LoRAs.
- Nano Banana
Fast, design-forward AI image generator — strong aesthetic quality, used inside Kimi and Manus for slide visuals.
- Stable Diffusion
Stability AI's open image models — the backbone of much open-source gen.