Google I/O 2026 was full of exciting announcements. Buried underneath them was a change that landed in millions of inboxes on May 20 with far less fanfare: Google AI Pro is moving to a compute-based usage system where limits depend on prompt complexity, chat length, and the AI tools being used — instead of fixed quotas. The 1,000 monthly AI credits previously included in the base plan are gone. And early user reactions are largely negative.
Table of Contents
What Exactly Changed
Starting May 20, 2026, Google is introducing compute-based usage limits in the Gemini app that factor in the complexity of your prompt, the features you use, and the length of your chat. Your limit refreshes every 5 hours until you reach your weekly limit. As an AI Pro subscriber, you enjoy a 4x higher usage limit than non-subscribers.
The product-based usage limit model is also rolling out to other products, starting with Flow and Antigravity. While 1,000 AI credits will no longer be included as a benefit in your base plan each month, Google says the new usage limit model should allow you to maintain the same experience you are used to.
The tasks that drain quota fastest are now confirmed: image, video, and music generation; Deep Research; the Pro Model; and extended thinking features including Deep Think all draw heaviest on usage limits.
The New Subscription Tier Structure
Google has restructured its AI subscription tiers as follows:
| Plan | Price | Usage Limit | Storage | Key Features |
|---|---|---|---|---|
| Google AI Plus | $7.99/month | 2x free tier | 200GB | Gemini, basic access |
| Google AI Pro | $19.99/month | 4x free tier | 5TB | Pro model, YouTube Premium Lite |
| Google AI Ultra (New) | $99.99/month | 5x Pro tier | 20TB | Gemini Spark, Project Genie |
| Google AI Ultra (Top) | $200/month (was $250) | 20x Pro tier | 20TB | All features, max limits |
If you burn through your quota on bigger models, Google automatically shifts you to smaller, faster models so you never lose access entirely. AI Pro and Ultra subscribers can buy pay-as-you-go top-up AI credits for Google Antigravity, Google Flow, and soon the Gemini app itself.
What Is Not Changing
To be fair to Google, the fundamentals remain intact. Model access continues — you still get the latest Gemini Flash and Pro models with Thinking capabilities. Features including Deep Research, video generation, and more remain available. And 5TB of shareable storage across Google Photos, Drive, and Gmail is unchanged.
Google’s stated rationale is reasonable on paper: a simple text prompt uses far less compute than a complex video or coding prompt, so treating them equally was never actually fair to light users who subsidised heavy ones.
Why Users Are Frustrated
A user on Reddit called the new system “totally scam” after a single text prompt consumed 13% of their quota. That is the core problem: the limits are not transparent enough for users to predict their consumption before they hit a wall.
Early reactions to these limits are mostly opposed to the change. Google previously did not directly advertise actual figures for the difference in usage limits on its AI plans — instead using words like “more” and “higher” — so the new system, while more honest in theory, lands harder in practice because users have no clear benchmark to plan against.
The five-hour refresh window is also a double-edged sword. On one hand it prevents a day-long lockout from a single heavy session. On the other, Google is following Anthropic’s approach of using five-hour windows to restrict usage — a model that power users on any platform find less predictable than a simple daily reset.
The Bigger Picture: AI Is Moving to Compute Billing
This is not a Google-specific story. It is an industry inflection point.
Flat-rate “unlimited” AI subscriptions made sense when usage patterns were predictable and model inference was cheap. In 2026 — where users are running Gemini Omni video generation, Deep Research sessions, and multi-step Antigravity agent workflows from the same subscription — the economics of flat-rate pricing no longer hold.
Subscription AI providers are increasingly shifting to compute-tied metering, creating less predictable per-request costs for teams. Google is not the first — OpenAI’s tiered usage, Anthropic’s Claude rate limits, and now Google’s compute credits all follow the same direction of travel. The AI subscription era of simple unlimited access is ending. What replaces it looks a lot more like cloud billing — and for power users, that means costs are going up whether or not the headline subscription price changes.
What You Should Do Now
If you are on Google AI Pro, check your usage in Gemini Settings → Usage Limit to understand your baseline consumption before you hit a wall mid-workflow. Reserve Deep Research, video generation, and extended thinking for tasks that genuinely need them — basic text queries, summarisation, and drafting cost almost nothing under the new model.
For Indian users evaluating AI subscriptions, Claude Pro at $20/month and ChatGPT Plus at $20/month follow similar consumption-based patterns — so the comparison is less about which platform is “unlimited” and more about which one’s compute budget goes furthest for your specific use case.
The Bottom Line
The new quota system has drawn criticism from users on Reddit and social media, with many arguing that the updated limits feel less transparent and harder to predict compared to the earlier usage model. Google’s rationale is sound — heavy compute tasks should cost more than light ones. The execution needs work: clearer per-task cost estimates and a real-time usage dashboard would go a long way toward defusing the frustration. Until then, this is the new reality of AI subscriptions — compute billing is here, and it is only going to get more granular.
Stay tuned to TechnoSports for the latest AI subscription updates, Google I/O 2026 coverage, and AI tools news in India.
