We have partnered with Black Forest Labs (BFL) again to bring their optimized FLUX.2 [klein] 9B model to Workers AI. This distilled model offers enhanced quality compared to the 4B variant, while maintaining cost-effective pricing. With a fixed 4-step inference process, Klein 9B is ideal for rapid prototyping and real-time applications where both speed and quality matter.<\/p>\n
Read the BFL blog<\/a> to learn more about the model itself, or try it out yourself on our multi modal playground<\/a>.<\/p>\n Pricing documentation is available on the model page<\/a> or pricing page<\/a>.<\/p>\n The model hosted on Workers AI is optimized for speed with a fixed 4-step inference process<\/strong> and supports up to 4 image inputs. Since this is a distilled model, the With the REST API, the multipart form data input looks like this:<\/p>\n With the Workers AI binding, you can use it as such:<\/p>\n The parameters you can send to the model are detailed here:<\/p>\n Required Parameters<\/strong><\/p>\n Optional Parameters<\/strong><\/p>\n Note:<\/strong> Since this is a distilled model, the The FLUX.2 klein-9b model supports generating images based on reference images, just like FLUX.2 [dev] and FLUX.2 [klein] 4B. You can use this feature to apply the style of one image to another, add a new character to an image, or iterate on past generated images. You would use it with the same multipart form data structure, with the input images in binary. The model supports up to 4 input images.<\/p>\n For the prompt, you can reference the images based on the index, like You must name the input parameter as Workers AI platform specifics<\/h4>\n
steps<\/code> parameter is fixed at 4 and cannot be adjusted. Like FLUX.2 [dev] and FLUX.2 [klein] 4B, this image model uses multipart form data inputs, even if you just have a prompt.<\/p>\nJSON Schema for Model<\/summary>\n
\n
prompt<\/code> (string) – Text description of the image to generate<\/li>\n<\/ul>\n\n
input_image_0<\/code> (string) – Binary image<\/li>\ninput_image_1<\/code> (string) – Binary image<\/li>\ninput_image_2<\/code> (string) – Binary image<\/li>\ninput_image_3<\/code> (string) – Binary image<\/li>\nguidance<\/code> (float) – Guidance scale for generation. Higher values follow the prompt more closely<\/li>\nwidth<\/code> (integer) – Width of the image, default 1024<\/code> Range: 256-1920<\/li>\nheight<\/code> (integer) – Height of the image, default 768<\/code> Range: 256-1920<\/li>\nseed<\/code> (integer) – Seed for reproducibility<\/li>\n<\/ul>\nsteps<\/code> parameter is fixed at 4 and cannot be adjusted.<\/p>\n<\/details>\nMulti-reference images<\/h4>\n
take the subject of image 1 and style it like image 0<\/code> or even use natural language like place the dog beside the woman<\/code>.<\/p>\ninput_image_0<\/code>, input_image_1<\/code>, input_image_2<\/code>, input_image_3<\/code> for it to work correctly. All input images must be smaller than 512×512.<\/p>\n