Runway: Gen-4.5

Runway: Gen-4.5 is a video generation model from Runway, the company behind the Gen family of generative video tools. It sits in the video category and is built for cinematic scene creation, with Runway describing its strengths as motion quality, visual fidelity and prompt adherence. If you already know Runway's model line, this is the Gen-4.5 entry, and on api14 it is addressed by the model id runway/gen-4.5.

The model accepts text and image inputs and produces video as its only output. Text prompts drive text-to-video generation, while an image can be supplied as a first frame for image-to-video workflows, which lets you anchor a shot to a specific starting composition. Output is 720p, and you can choose between 16:9 and 9:16 aspect ratios, covering standard landscape and vertical formats. Clip durations run from 2 to 10 seconds in one second increments, so you can request anything from a very short beat to a ten second shot. Audio generation is not available, so plan to add sound separately. No context length is listed for this model, which is expected for a video generator rather than a text model.

The practical uses follow from those capabilities. Text-to-video suits concept work, storyboards that need to move, and short establishing shots where you want to iterate on a prompt quickly. Image-to-video is useful when you have a still you want to animate, such as a product photo, a keyframe from an artist, or a frame pulled from an existing edit, and you want the motion to start from that exact image. The 9:16 option makes the model relevant for vertical formats, and the 2 to 10 second range fits social clips, ads and inserts rather than long-form sequences. Because there is no audio track, it works best as a visual source that you assemble in an editor.

You call it through api14's OpenAI-compatible API using the model id runway/gen-4.5, so existing OpenAI-style client code works with a change of base URL and model name. The model exposes no tunable request parameters beyond the model id itself, so the prompt and any first frame image carry the creative direction. Duration, aspect ratio and resolution are fixed by the model's supported set rather than exposed as adjustable parameters, so you should not expect to pass sampling or generation settings through the request.

If you need audio generated alongside the video, or resolutions above 720p, this model does not provide them and another option may fit better. It is also a poor match for long continuous footage, since the maximum clip is 10 seconds. For still image generation or text tasks, a different model on api14 is the right choice.