Google cuts video AI costs in half with Veo 3.1 Lite

Running AI video in production has always felt like a luxury hobby for developers. If you wanted to build an automated pipeline or generate user clips at scale, the computing bill alone was enough to kill the project before launch.

Google is tackling that cost wall head-on with Veo 3.1 Lite, its new budget-friendly video generation model. It is rolling out today in paid preview across the Gemini API, Google AI Studio, and Vertex AI.

Half the price, same rendering speed

The headline feature here is pure economics. Google claims Veo 3.1 Lite costs less than 50% of what it charges for Veo 3.1 Fast, while matching the exact same generation speed.

It handles both Text-to-Video and Image-to-Video workflows. You can generate in landscape 16:9 or vertical 9:16 framing, with output options set at 720p and 1080p.

Developers can also choose generation lengths of 4, 6, or 8 seconds, with pricing scaling based on the duration you pick. To sweeten the deal, Google announced it will also slash prices for Veo 3.1 Fast starting April 7.

A complete tier stack and a universal upscaler

With Lite entering the picture, the Veo 3.1 family on Vertex AI now has a clear three-tier structure, and all three models support native audio generation.

The standard Veo 3.1 handles top-tier visual fidelity for master cuts, Veo 3.1 Fast tackles regular production workflows, and Veo 3.1 Lite focuses strictly on high-volume apps and rapid prototyping.

Alongside the model, Google also introduced a standalone Veo upscaling tool on Vertex AI, currently in private preview with a public preview coming soon.

It can upscale low-resolution footage up to 1080p and 4K. What makes it neat is that it is model-agnostic: you can feed it footage from Veo, clips from rival AI models, or even grainy footage captured on an old physical camera.

Why it matters

For indie builders and startups, Veo 3.1 Lite finally makes automated, high-volume video workflows financially viable. Getting the pricing under half the cost of Fast means you can actually let users experiment without fearing a catastrophic cloud bill at the end of the month.

We think Google’s real masterstroke here is the pairing of Lite with the new universal upscaler. Instead of paying premium rates to render high-resolution 4K clips from scratch, a smart pipeline can now generate dozens of cheap 720p candidates using Lite, pick the best one, and upscale it on demand.

This puts direct pricing pressure on standalone players like Runway and Luma. When raw model intelligence starts hitting diminishing returns, the cloud giant with the deepest infrastructure discounts usually wins. The only real question now is how much physical consistency Lite had to sacrifice under the hood to get that cheap.