Google released Veo 3.1 Lite, which directly reduces the price of AI video generation to US$0.05 per second. For developers, this is not a replacement for ordinary models, but an increase in the trial and error cost of Wensheng videos and graphic videos to a range where batches can be iterated: generating an 8-second 720p video for only US$0.40, and it also comes with its own environment. Sound effects and sound synchronization.
The price is not just lower
According to Google's current public price, Veo 3.1 Lite is US$0.05 per second at 720p and US$0.08 per second at 1080p. For comparison, Veo 3.1 Fast currently costs $0.15 per second for 720p and 1080p, while Veo 3.1 Standard costs $0.40 per second.
This means that the same 8-second 720p video costs $0.40 for Lite,$1.20 for Fast, and $3.20 for Standard. For people who occasionally do a few samples, this is just a smaller bill; for teams that have to run a lot of A/B testing, batch release of material, and automate video workflows, it directly changes the product model.
Google also predicts that Veo 3.1 Fast will continue to reduce prices on April 7. But even at current prices, Lite has shifted the matter of "making a few more editions and seeing" from careful trial to a development action that can be performed on a daily basis. AI video generation used to be stuck between quality and cost, and Lite is more like removing the cost threshold first.
Lite is not a cheap silent version
Cheap does not mean that it can only provide basic functions. Veo 3.1 Lite still supports text-generated video and image-generated video, supports 16:9 horizontal screen and 9:16 vertical screen, and also supports three lengths of 4 seconds, 6 seconds, and 8 seconds, covering 720p and 1080p output. This set of specifications is enough for Short Video platforms, advertising materials and in-app dynamic content.
More importantly, Lite still retains one of the most recognizable capabilities of the Veo 3.1 family: native video with audio. Many low-cost models can make the picture first, but leave the sound to later splicing, making the rhythm, mouth shape and atmosphere easy to disconnect. Google did not remove the audio this time, indicating that Lite's goal is not a "castrated demonstration model", but a production model for large-scale calls.
Of course, Lite is not an all-around player. Public documents show that it does not support 4K or Extension, and 1080p is limited to 8 seconds. Google's hierarchy is clear: Lite is responsible for high frequency, low cost and large scale; higher specifications and stronger control are still reserved for other Veo 3.1 gears.
What do developers really buy?
As the AI video track reaches 2026, the focus of competition is not just "who is more like a movie", but "who can enter the business process." For developers and corporate customers, the most expensive thing is never a single video, but an iteration. An e-commerce App wants to generate short product videos in batches, a game team wants to quickly produce character promotional videos, and a marketing platform wants to automatically generate portrait advertisements. Every time the unit cost drops, the product boundary will expand.
The value of Veo 3.1 Lite is falling on this line. It advances high-quality video generation from a small number of high-quality trials to high-frequency experiments, template production and large-scale distribution. Especially when 9:16 vertical screen and short content within 8 seconds has become the mainstream distribution format, the cost performance of 720p and 1080p is closer to real market demand than simply pursuing 4K.
Google is completing the video model gradient
From Standard to Fast to Lite, Google is making Veo 3.1 a more complete price band. Standard is responsible for high quality and higher ceilings, Fast is responsible for speed and intermediate costs, and Lite pulls video generation into a budget range that makes it easier to access on a large scale. For the Gemini API and Google AI Studio, this also makes Veo no longer just a flagship model to showcase capabilities, but more like an infrastructure that can be embedded into products, services and workflows.
The next stage of AI video is not necessarily determined by the quality ceiling, but more likely by the cost of generating each video. Whoever first changes "can do" to "can do" will have a better chance of receiving the developer ecosystem and corporate calls. Veo 3.1 Lite may not be Google's strongest video model, but it may well be the card it comes closest to reaching scale.
