Krea: Krea 2 Large

Krea: Krea 2 Large is an image generation model from Krea. It sits at the top of the Krea 2 line, more than twice the size of Krea 2 Medium, and its lighter post-training gives the images it produces a rawer, more textured and more flexible character than you might expect from a model tuned heavily for polish. If you are choosing an image model for a project, this is the larger, higher-capability option in the family.

On the input side it accepts text and images, and it returns images. It has a context window of 65,536 tokens and a maximum output of 58,982. Generation is controlled through a small set of image parameters rather than a broad sampling API. Resolution is fixed to a single enum value, 1K, so you cannot request other sizes. Aspect ratio is selectable from 1:1, 4:3, 3:2, 16:9, 4:5, 2:3 and 9:16. You can supply between zero and one input reference images, and there is a seed option exposed as a boolean.

Those capabilities point to a fairly specific set of uses. Text to image work is the obvious starting point, including layout driven tasks where you match an aspect ratio to a target canvas, such as 16:9 for wide banners or 9:16 for vertical formats. Because the model accepts an image as well as text, you can also pass a single reference image alongside your prompt, which suits work where you want to guide style, composition or subject while still generating a new image. The 1K resolution keeps outputs in a range that is practical for web and preview use, and the seed control helps when you want to reproduce or iterate on a result.

You call the model through api14's OpenAI-compatible API using the model id "krea/krea-2-large". The request follows the same shape as other image requests on the gateway, so existing OpenAI-style client code needs little more than a change of model id. The model's supported parameter list is empty, meaning no general text generation parameters are advertised for it. Instead, generation is configured through the image parameters described above: resolution, aspect ratio, input references and seed. Text and image inputs go in, an image comes out.

It is worth being clear about the constraints before you commit. Output is image only, so if you need text back from the same call, this is not the model for that. Resolution is limited to 1K, so work that requires larger output needs a different option. And since no supported parameters are listed, any request that depends on tuning knobs like temperature or top_p will not have those applied here. Within those bounds, Krea 2 Large is a reasonable fit for reference-guided and text-driven image generation at 1K.