Gen-3 Alpha

Gen-3 Alpha is the text-to-video model Runway unveiled on June 17, 2024 as the successor to Gen-2 and the first of a new family of foundation models trained on new large-scale multimodal infrastructure. It generates 5- or 10-second 720p clips from text prompts, still images, or short video references, and was widely seen as a major step up from Gen-2 in fidelity, motion realism, and prompt adherence.

Runway extended it over 2024 with image-to-video (July 29, 2024), an Extend feature for stitching longer sequences, video-to-video, and the Act-One character-performance feature. A faster sibling, Gen-3 Alpha Turbo, followed at roughly seven times the speed and half the per-second credit cost. Gen-4 has since superseded it as Runway's flagship. The honest picture: output quality is genuinely strong, but the credit math is brutal for iteration — a 10-second clip burns about 100 credits, and extensions cost the same per-second rate as the original generation, so experimenting eats a monthly allowance fast. It is proprietary and available only through Runway's paid plans.