Runway Introduces Gen-4.5 Model for Faster, Longer Text-to-Video Creation

Details

• Runway has launched Gen-4.5, its latest text-to-video model, now available across its web studio, mobile app, and public API.
• The new version allows users to create up to 15-second clips at 1080p and 24 fps, while halving the average rendering time compared to Gen-4.
• Gen-4.5 adds a "World Control" panel, enabling users to lock camera paths, depth maps, or reference frames for consistent spatial movement in multi-shot videos.
• The model was trained using a blend of licensed and synthetic footage and passed Runway’s internal safety and copyright checks for commercial use.
• Gen-4.5 is now the standard for all subscription tiers, with Gen-4 API calls auto-upgrading by January 2026.
• Runway plans to release a research paper in early 2026 detailing the technical advancements and data sources for Gen-4.5.

Impact

This release ups the ante for competitors like OpenAI’s Sora beta and Pika Labs’ Video V5, which lag behind in clip duration and features. Faster generation and improved controls are set to benefit content creators and enterprises while ensuring compliance with new data provenance and copyright rules. Runway’s fast update cycle and expanded options could spur rivals to quicken innovation in the rapidly evolving AI video space.