Google’s Veo 3.1 Lite
Google released Veo 3.1 Lite, a lower‑cost, high‑speed video generation option exposed through the Gemini API for developers who need embeddable video features without massive compute bills. The release is pitched as a pragmatic alternative for startups needing scalable generative video after bigger players retrench. (marktechpost.com)
Google published the Veo 3.1 Lite announcement on March 31, 2026 and made the model available in a paid preview via the Gemini API and Google AI Studio, with the blog post credited to product manager Alisa Fortin and developer advocate Guillaume Vernade. (blog.google) Google states Veo 3.1 Lite runs at the same generation speed as Veo 3.1 Fast while costing under 50% of Fast, and the company said it will lower Veo 3.1 Fast pricing on April 7, 2026. (blog.google) Official specs list Text-to-Video and Image-to-Video inputs, output video with audio, two resolutions (720p and 1080p), native landscape (16:9) and portrait (9:16) framing, and fixed clip durations of 4, 6, or 8 seconds. (blog.google) Google’s Gemini API model page and preview documentation note Veo 3.1 Lite does not support 4K outputs or Scene Extension and documents input/output limits such as a 1,024-token text input ceiling. (ai.google.dev) DeepMind’s Veo technical report and multiple previews describe the family’s architecture as a latent diffusion system with a transformer‑based denoising network that operates on spatio‑temporal latents to improve temporal consistency. (storage.googleapis.com) The Veo 3.1 Lite rollout arrived days after OpenAI announced on March 24, 2026 that it would discontinue the Sora app (web/app experiences ending April 26, 2026 and the Sora API on September 24, 2026), a timing observers have linked to a market shake‑up in developer-facing video APIs. (help.openai.com)
