Google's AI Video Stack in 2026: Veo 3.1 Lite Pricing, Upscaling on Vertex AI, and the New Cost Baseline

News

Google Launches Veo 3.1 Lite: Most Cost-Effective Video Model Now on Gemini API

Veo 3.1 Lite is now live on the Gemini API at under 50% of Veo 3.1 Fast's cost, with the same generation speed — a meaningful shift for developers building high-volume video pipelines. Support for text-to-video, image-to-video, 720p/1080p, and 4/6/8-second durations makes it a flexible drop-in for production workflows. blog.google

Analysis

Veo 3.1 Lite Technical Breakdown: DiT Architecture, SynthID Watermarking, and the $0.05/sec Cost Floor

This deep-dive covers the Diffusion Transformer backbone powering Veo 3.1 Lite, including how spatio-temporal patches maintain temporal consistency without blowing up compute costs. The inclusion of mandatory SynthID watermarking is worth noting for any enterprise use case where synthetic media provenance matters. marktechpost.com

Veo 3.1 Family Now Complete: How the Three-Tier Model Stack Compares for Different Use Cases

With Veo 3.1 Lite rounding out the family, Google now offers a full three-tier stack — Lite for high-volume iteration, Fast for standard production, and flagship Veo 3.1 for top-fidelity final cuts — all with native audio generation. This tiered structure gives developers a clear decision framework rather than a one-size-fits-all model. 9to5google.com