Google doubles AI credits for creative tools Flow and Whisk

If you have ever tried making a complete video with AI, you know the real bottleneck isn't imagination—it's watching your generation credits disappear after ten botched generations.

Google Labs is giving creators some breathing room. The team just announced that it is doubling the monthly AI credits for Google AI Ultra subscribers, jumping from 12,500 to 25,000 credits.

100 million videos later, Google pays the bill

The move comes on the heels of a massive milestone for Flow, Google's AI filmmaking tool. According to Google, users have generated over 100 million videos in Flow since it launched in May, ranging from personal clips and cinematic shorts to branded commercial work.

With creators burning through compute to piece together coherent scenes, Google is essentially loosening the leash so people can experiment without counting pennies on every render.

These expanded credits aren't restricted to Flow. Subscribers can also spend them in Whisk, Google's experimental tool that generates images using both text and image prompts. Whisk is also kicking off a major international expansion, rolling out to 77 more countries starting tomorrow.

Existing Google AI Ultra subscribers will see the 25,000 credit allocation kick in on their next billing renewal, while Workspace users will see the update roll out over the coming days.

Why it matters

For digital creators and visual storytellers, this makes Google AI Ultra significantly more attractive right now. Filmmaking with generative AI requires constant iteration, and doubling the credit cap effectively cuts the cost of failure in half. You can afford to re-roll that stubborn camera pan.

From a competitive angle, we think Google is playing smart defense. As video generation models proliferate across the market, platform lock-in comes down to who provides the most frictionless creative suite. Linking Flow and Whisk under a generous shared credit pool keeps creators building inside Google's garden instead of jumping across single-purpose web apps.

That said, throwing more tokens at the problem highlights an underlying reality: generative video is still wildly unpredictable. Doubling prompt allowances solves the symptom of frequent rerolls, but until models offer deterministic camera and character consistency, creators will keep burning through whatever limit Google sets.