Sourceful: Riverflow V2 Fast

Sourceful: Riverflow V2 Fast is an image generation model from Sourceful, part of the company's Riverflow 2.0 lineup. Sourceful positions it as the fastest variant in that series, aimed at production deployments and latency-critical workflows. It sits in the image category and is intended for developers who need generated images returned quickly rather than for offline or batch experimentation where speed matters less.

The model accepts text and image inputs and produces images as its only output. Its context window is 8192 tokens, with a maximum output of 7372 tokens. Image generation is controlled through a set of image parameters rather than the usual text generation knobs. Resolution is an enum with the values 1K, 2K and 4K. Aspect ratio is an enum with the values 1:1, 4:3, 3:4, 3:2, 2:3, 16:9, 9:16, 21:9 and auto. The n parameter is a range from 1 to 1, so each request returns a single image. Input references are a range from 0 to 4, which means you can attach up to four reference images to guide the output, or none at all for pure text to image work.

Those capabilities point to a few practical uses. Text to image generation covers app features such as avatars, thumbnails, product mockups and illustrative assets, where the resolution and aspect ratio enums let you match the target layout directly. Because image inputs are accepted alongside text, the model also fits reference guided workflows: passing one to four input references lets you steer style, composition or subject while still generating a fresh image. The 8192 token context window bounds how much text and how many references you can supply in one call, so prompts and reference sets should stay within that budget. The "fast" designation makes this variant the sensible pick when a user is waiting on the result, for example in an interactive editor or a request path with a tight timeout, rather than for jobs where you would rather trade latency for other considerations.

You call the model through api14's OpenAI-compatible API using the model id "sourceful/riverflow-v2-fast". Because the surface is OpenAI-compatible, you point your existing OpenAI client or HTTP requests at api14's endpoint and set that model id in the request. The model exposes no supported request parameters in the usual sense; instead, generation is configured through the image parameters described above (resolution, aspect ratio, n and input references). There is no separate list of sampling parameters such as temperature or top_p to tune here, so the meaningful controls are the image ones.

Keep the limits in mind when choosing it. The context window of 8192 tokens and the maximum output of 7372 tokens cap how much text and how many input references fit in a single request, and n is fixed at 1, so you cannot request multiple images in one call. If your workload depends on tunable text generation parameters, this model does not provide them. For those cases, or when latency is not the constraint, another model in the Riverflow 2.0 lineup or a different image model may fit better.