Flux 2 Klein is a dual-model AI image generator that produces high-quality images from text prompts or reference images, offering a choice between a fast distilled 4-billion parameter model and a detailed 9-billion parameter base model.
What is Flux 2 Klein?
Flux 2 Klein is an open-source dual-model image generation suite developed by Black Forest Labs. It takes text prompts or image inputs and outputs high-resolution PNG images. The suite includes Flux 2 Klein 4b (Distilled) & 9b Base — a 4-billion parameter distilled model optimized for speed and a 9-billion parameter foundation model built for maximum detail. Both run on cloud infrastructure via a web generator, supporting text-to-image and image-to-image workflows with native aspect ratios (1:1, 16:9, 9:16, etc.).
Key Features
- Dual-model architecture — Choose between Flux 2 Klein 4b (distilled, ~400M params equivalent in speed) for fast iterations or Flux 2 Klein 9b Base for highest quality and spatial reasoning. Both use a rectified flow transformer for efficient sampling.
- Text-to-Image & Image-to-Image — Generate images from a text prompt (up to 2000 characters) or remix an existing image using a reference image input. Output format is PNG.
- Multi-reference consistency — Supply up to 5 reference images to guide character appearance, style, or composition across generations.
- Advanced text rendering — Produces coherent, readable text inside images (e.g., posters, book covers, logos) without the typography artifacts common in older models.
- Native aspect ratios — Supports 1:1, 16:9, 9:16, and other ratios without stretching or cropping artifacts, thanks to multisize training.
- Safety fine-tuning — The models include built-in safety measures to reduce harmful content generation, making them suitable for public-facing applications.
Who is it for?
- UI/UX designers — Rapidly prototype interface mockups, landing pages, and icon sets with consistent styling, using the 4b model for speed and the 9b model for final polish.
- Photographers and digital artists — Generate photorealistic product shots, lighting studies, or compositional sketches with the 9b Base model’s superior detail and lighting accuracy.
- Generative AI developers — Fine-tune the 9b base model for specialized tasks (e.g., character consistency, brand assets) using its full training signal and flexible conditioning.
What can you do with Flux 2 Klein?
- Iterative concept design: Use the 4b model to generate dozens of mood board images in minutes, then switch to the 9b model to refine the best candidates with higher detail.
- Consistent character art: Supply multiple reference images of a character (up to 5) and generate new poses, expressions, or scenes while maintaining identity.
- Typography-heavy graphics: Create posters, book covers, and logo concepts with legible text — a feature that distinguishes Flux 2 Klein from many open-source text-to-image models.
How does Flux 2 Klein work?
Select a model (Klein Fast 4b or Klein Pro 9b), choose an aspect ratio and output format (PNG), write a prompt (up to 2000 characters), and click “Generate” (costs 1 credit per generation). The image result appears in the preview area. An optional image-to-image mode lets you upload a starting image to guide the output.
Pricing
Freemium — generation costs credits; exact credit pricing and free quota are not listed on the product page. Start Generating to see current terms.
