Sourceful: Riverflow V2 Pro

Sourceful: Riverflow V2 Pro is an image generation model from Sourceful, and it sits at the top of the company's Riverflow 2.0 lineup. Sourceful positions it as the most powerful variant in that series, aimed at users who need top-tier control over the generated result and accurate text rendering inside images. It is an image model rather than a chat or text model, so its role on api14 is to turn prompts and reference material into pictures.

The model accepts text and images as input and returns images as output. Its context window is 8,192 tokens, with a maximum output of 7,372, which covers the prompt and any reference images you supply along with the generation request. Output resolution is chosen from three options: 1K, 2K or 4K. Aspect ratio is also selectable, with 1:1, 4:3, 3:4, 3:2, 2:3, 16:9, 9:16, 21:9 and auto available. A single request produces exactly one image, since the n parameter has a minimum and maximum of 1, and you can attach up to 10 input references to guide the result.

Those capabilities point to work where fidelity and layout matter. Because the model handles image inputs alongside text, it suits reference-driven generation, where you supply one or more existing images and ask for a variation, a restyle or a new composition that follows them. The high resolution ceiling makes it a reasonable fit for assets that will be viewed or printed at size, such as marketing visuals, product imagery, posters and social formats, and the range of aspect ratios means you can target square, portrait, landscape and ultrawide placements without cropping afterwards. Sourceful's emphasis on text rendering suggests it is also worth trying for images that must contain readable words, such as banners or promotional graphics, though you should verify results on your own content.

Calling it through api14 works the same way as any other model on the gateway, using the OpenAI-compatible API with the model id "sourceful/riverflow-v2-pro". You send your prompt as text and any reference images as input, then read the generated image from the response. The model exposes no general supported parameters, so sampling controls such as temperature or top_p are not part of its interface. Instead, generation is steered through image-specific options: resolution, aspect_ratio, n (fixed at 1) and input_references (0 to 10).

Keep the constraints in mind when choosing it. It outputs images only, so it cannot be used for text completion, chat or reasoning tasks, and it will not return multiple images in one call. The 8,192 token context is modest compared with large language models, which matters if you plan to pass many or very detailed references. If your task is text generation or analysis rather than image creation, a text model on api14 is the better fit.