Editorial illustration for Gemini Omni adds AI video generation, using compute limits based on complexity and size
Gemini Omni adds AI video generation, using compute...
Google's latest Gemini model can now make videos. Not well, but that's beside the point. The important thing is the mechanism: a new, fluid system of rationing compute power that changes with each request.
You get a budget. How much you spend depends on what you ask for. A simple, short clip costs less.
A complex, longer sequence drains your account faster. This isn't a flat rate. It's a meter, running.
The results are predictably mixed. Feed it a prompt or an image and it will generate a sequence. Sometimes quickly.
Sometimes in a style that vaguely matches your request. The output is short, stamped with a watermark, and locked down by regional and content filters. This is a controlled demo, not a tool.
The feature itself is secondary. Google is testing a new economic model for generative AI. One where your usage isn't measured in simple queries, but in computational weight.
It's a glimpse of the infrastructure being built beneath the flashy demos. The videos are rough drafts. The billing system is the final product.
Common Questions Answered
How does Gemini Omni's compute budget system work for video generation?
Gemini Omni uses a dynamic compute rationing system where users receive a budget that fluctuates based on the complexity and length of the requested video. Simple, short clips consume less budget, while complex, longer sequences drain the account faster, creating a metered billing approach rather than a flat-rate model.
What factors determine how much compute power is spent on a Gemini Omni video request?
The computational cost depends on the complexity and size of the video being generated. Users can input either text prompts or images to generate videos, and the system calculates the required compute resources based on these input parameters and the desired output specifications.
Why is Gemini Omni's billing mechanism more significant than its video generation capability?
Google is using Gemini Omni's video generation feature to test a new economic model for generative AI that measures usage by computational weight rather than simple query counts. This billing system represents the infrastructure being built for future AI services, making it more important than the current video quality, which Google acknowledges is mixed.
What does Google's new computational weight-based billing model mean for generative AI pricing?
Instead of charging per query or request, Google's model charges based on the actual computational resources required for each task. This approach allows for more granular and accurate pricing that reflects the true resource consumption, moving away from traditional flat-rate or per-query billing structures used in earlier generative AI systems.
Further Reading
- Gemini Apps limits & upgrades for Google AI subscribers — Google Support
- Gemini Omni – Create & edit videos as easy as having a conversation — Google Gemini
- Google Gemini AI New Limits Based on Compute - AI is Expensive — YouTube
- Gemini Usage Limits Explained : Never Run Out Again — YouTube
- Is Gemini Omni Free? Plans, Limits, Credits & Access — Veo3 AI