Recraft: Recraft V4.1 Pro Vector

Recraft V4.1 Pro Vector is an image generation model from Recraft, the vector (SVG) variant of Recraft V4.1 Pro. It is tuned for high aesthetics, which means the visual quality of the output is a deliberate focus of this variant rather than an afterthought. If you already know Recraft's line of image models, this is the one aimed at work that needs to end up as scalable vector artwork rather than a raster image.

The model accepts text and image inputs and produces image output, specifically SVG. It has a context length of 65,536 tokens and a maximum output of 58,982. The image parameters give you control over the shape and quantity of what comes back: aspect_ratio accepts 1:1, 4:3, 3:4, 16:9, 9:16 or auto, output_format accepts svg, n accepts a value from 1 to 6, and input_references accepts 0 or 1 reference image. The provider describes the output as higher-resolution SVG across the supported range.

Those capabilities point to a fairly specific set of jobs. Because the output is SVG, the results stay editable and scale cleanly, which suits logos, icons, illustrations, diagrams and other assets that need to hold up at different sizes or be adjusted later in a vector editor. The single input reference slot is useful when you want the generated artwork to follow a supplied image, for example matching a style or building on an existing mark. Generating up to six variations in one request helps when you are exploring directions before committing to one. Text input covers the usual prompt-driven generation, and the combination of text plus one reference image covers the case where a written description alone is not precise enough.

You call the model through api14's OpenAI-compatible API using the model id "recraft/recraft-v4.1-pro-vector". Because the interface is OpenAI-compatible, you can point existing OpenAI client libraries at api14 and change the model id rather than rewriting your integration. Note that the model lists no supported parameters in the usual chat sense; the controls it exposes are the image parameters described above (aspect_ratio, output_format, n and input_references), so structure your requests around those rather than around sampling or token-level options.

It is worth being clear about what this model is not. It outputs SVG only, so if you need a raster format such as PNG or JPEG, this variant will not produce it. It is also an image model, not a text or chat model, so tasks like summarisation, reasoning or code generation fall outside its scope and belong with a text model instead. Within vector image generation, though, the combination of text and image inputs, a single reference image and up to six outputs per request covers a lot of ground.