Breaking

Google Overhauls Gemini Usage Limits, Moving From Daily Prompt Counts to Compute-Based Quotas

July 18, 2026 at 06:06 EDT

Foundation Models
Sales & Business

Google has replaced the simple daily prompt caps in its Gemini apps with a compute-based system, a change unveiled at Google I/O 2026 that means heavy tasks such as video generation and deep research now eat into a user's allowance far faster than a quick text query. The shift, announced around May 19, 2026, marks one of the most consequential adjustments to how consumers access Google's flagship AI assistant since its launch.

May 19, 2026 · Google I/O 2026

Gemini Ditches Daily Prompt Caps for a Compute-Based Meter

Limits now scale with a prompt's complexity — video, coding and Deep Research burn allowance far faster than a quick text query. Quotas refresh every five hours under a weekly ceiling.

5 hrs

rolling window before allowance refreshes, under a weekly ceiling

1–2 hrs

how fast Pro users hit ceilings on complex tasks at first

$100 / $200

new monthly AI Ultra pricing, bundled with perks

Allowance multipliers by tier

Relative to the standard allowance (×1)

×1

Standard

×2

AI Plus

×4

AI Pro

×20–80*

AI Ultra

*Ultra is reported at 5×–20× the Pro level; heights shown are approximate. Multipliers stratify how much compute allowance each tier receives.

When the weekly ceiling is hit

Larger, capable model

full quality

→

Falls back to smaller model

faster, reduced quality

→

Responses keep flowing

or buy top-up credits

The backlash

Pro subscribers hit their ceilings within an hour or two on video and Antigravity tasks, called it a major downgrade, and weighed cancelling or moving to Ultra.

Google's response

By late May, quotas were revised: a cap on how much any single request can consume, and failed requests no longer counting against the allowance.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.