Krea: Krea 2 Medium Turbo

Krea: Krea 2 Medium Turbo is an image generation model from Krea. It is a distilled, speed-focused variant of Krea 2 Medium, built for rapid iteration and graphic design exploration where fast generation matters more than the slower, heavier behaviour of a full model. If your team already knows the Krea family, this is the version to reach for when you want to move quickly through many visual ideas rather than produce a single carefully tuned render.

The model accepts text and image inputs and returns images. Its context window is 65,536 tokens, with a maximum output of 58,982. Image output is fixed at 1K resolution, and you can choose from seven aspect ratios: 1:1, 4:3, 3:2, 16:9, 4:5, 2:3 and 9:16. You can supply up to one input reference image, which makes it possible to steer a generation from an existing picture rather than text alone. A seed option is also available, so runs can be made reproducible when you need to compare variations or debug a prompt.

Those capabilities point to a fairly specific set of jobs. The speed focus and the aspect ratio range suit design exploration: generating many layout candidates for a poster, a social card or a vertical story format, then picking the strongest. The single reference image input is useful for style or composition transfer, for example feeding in a rough sketch or a mood board image and asking for variations. Because output is limited to 1K, the model fits web and screen mockups, thumbnails and concept passes rather than print-resolution final artwork.

You call it through api14's OpenAI-compatible API using the model id "krea/krea-2-medium-turbo". The model does not list any supported request parameters in the usual sampling sense, so the controls live in the image parameters instead: resolution (only 1K), aspect_ratio (the seven values above), input_references (zero or one reference image) and seed. If you are already sending requests to other models on api14, the request shape will look familiar, and you mainly need to set the image fields rather than temperature or top_p style options.

One limit worth noting before you commit: the fixed 1K resolution and the absence of listed sampling parameters mean this model is not the place to look for fine grained control over the generation process or for high resolution output. If your task needs larger images, or if you want a broader set of tunable parameters, another model in the catalogue may fit better. For fast, high volume visual iteration at 1K, Krea 2 Medium Turbo is the straightforward choice.