Gemini app adopts compute-based limits

Google changed how the Gemini app meters usage in mid-May, replacing prompt or message caps with a compute-based system that counts the complexity of a request, the features used and the length of a chat. The change took effect May 17, according to Google’s Gemini Apps Help pages. In the days that followed, users on social platforms and in community posts said they were hitting limits sooner than under the older system. Google has not published a simple public conversion table showing how many prompts a given request now consumes. ### When did Google make the change? Google’s Gemini Apps Help Center says “starting on May 17, 2026” the service would move to compute-based usage limits for personal-account users. The company says those limits determine how much a person can use Gemini tools and features by factoring in prompt complexity, the models and features used, and the length of the chat. 9to5Google and other outlets reported the change during the week of Google I/O, linking it to a broader overhaul of Gemini plan tiers and usage dashboards. Those reports said Google also began showing users more explicit limit information inside the app. ### What does “compute-based” mean in practice? Google says the new system is not based on a flat number of prompts. (support.google.com) Instead, heavier tasks consume more of a user’s allowance because the meter now reflects the work required to answer a request. The company’s language points to three variables: prompt complexity, feature use and chat length. Android Police reported that this brings Gemini closer to systems already used elsewhere in the industry, where a simple text exchange costs less than a longer session or a request that invokes more demanding tools. (9to5google.com) PCWorld described the shift as a move away from fixed request counts toward a model based on compute consumption. ### Why are some users saying their limit disappears faster? (support.google.com) 9to5Google reported on May 21 that early user reaction was largely negative, citing complaints that the new limits were less predictable and could be exhausted more quickly than the prior message-cap approach. The publication quoted Google’s notice to users saying limits refresh every five hours until a weekly limit is reached. (androidpolice.com) Google’s own support page helps explain those complaints, even without endorsing them. Because the meter now depends on complexity, two users sending the same number of prompts may consume very different amounts of allowance if one is using longer chats, more advanced models or heavier features. That is an inference from Google’s published description of the system. (9to5google.com) ### How do the new resets and paid tiers work? Google says Gemini app limits now refresh every five hours until a user reaches a weekly limit. The company also says paid subscribers receive higher ceilings than free users. On Google One plan pages, Google AI Plus is listed with “2x higher” Gemini usage limits than non-subscribers, Google AI Pro with “4x higher,” and Google AI Ultra with “up to 20x more limits” than the Pro plan. (support.google.com) Google’s pricing pages list Google AI Plus at $9.99 a month in the United States and Google AI Pro at $19.99 a month, while Ultra is marketed as starting at $100 a month. Those plan descriptions tie higher usage allowances directly to subscription level rather than to a published prompt count. ### Has Google adjusted limits after complaints? Android Headlines and 9to5Google reported on May 21 and May 22 that Google increased some Gemini usage quotas after user pushback, including for higher-end features tied to new plans. (support.google.com) Those reports indicate Google was still tuning the limits within days of the rollout. Google’s next public reference point is likely to remain its Gemini Apps Help page and Google One plan listings, which the company is using to describe resets, weekly caps and plan-specific multipliers. (one.google.com) (support.google.com) (androidheadlines.com)