Usage Limits Guide
Since May 2026, Gemini has no daily prompt counts. Google switched to compute-based limits that refresh every 5 hours inside a weekly ceiling, and publishes only multipliers: AI Plus is 2x standard limits, AI Pro is 4x, and AI Ultra is 5x–20x of Pro depending on tier. The concrete published numbers are the context windows — 32K tokens Free, 128K on Plus, 1M on Pro and Ultra — and heavier features (Deep Research, media generation, Deep Think) drain the allowance faster.
Figures marked derived are our arithmetic on official statements, not published numbers.
| What | Limit | Basis |
|---|---|---|
| Prompts | No published counts since May 2026 — compute-based, refreshing every 5 hours within a weekly ceilingofficial | official |
| Tier multipliers | AI Plus 2x standard · AI Pro 4x standard · AI Ultra 5x–20x of Proofficial | official |
| Context window | 32K tokens Free · 128K AI Plus · 1M AI Pro and Ultraofficial | official |
| Deep Research / Deep Think / media generation | Available by tier and "consume more of your usage" — no published countsofficial | official |
| Storage bundled | 5TB on AI Pro · 20TB on Ultra 5x · 30TB on Ultra 20xofficial | official |
Gemini 3.7 Flash
All tiers — the default fast model
Gemini Pro-class models
AI Pro and Ultra, at higher compute cost per prompt
Deep Think
AI Ultra tiers
Video generation (Veo/Omni)
AI Plus and above, heavy drain on the allowance
March 17, 2026
Google's four-tier lineup takes shape: AI Plus $7.99, AI Pro $19.99, AI Ultra 5x $99.99, AI Ultra 20x $199.99.
May 25, 2026
At I/O, Google replaces fixed daily prompt counts with compute-based limits refreshing every 5 hours inside a weekly ceiling — published limits become multipliers (Plus 2x, Pro 4x, Ultra 5x–20x of Pro).
Since May 2026, there is no published number. Google moved to compute-based limits that refresh every 5 hours within a weekly ceiling, and publishes only multipliers: AI Plus gets 2x standard limits, AI Pro 4x, and AI Ultra 5x–20x of Pro. Before the change, the fixed counts were roughly 5/day Free, 100/day Pro, 500/day Ultra.
AI Pro carries 4x the standard compute allowance and — the concrete part — a 1M-token context window against Free's 32K. In practice the context gap bites first: Free fails on long documents and codebases regardless of prompt budget. Pro also bundles Deep Research access and 5TB of storage at $19.99/mo.
Google no longer publishes a count. Deep Research draws from the same compute-based pool as everything else, just faster. Under the old fixed system AI Pro allowed up to 20 Deep Research reports per day, which remains a reasonable ceiling estimate for the current allowance.
Ultra comes in two tiers — 5x Pro limits at $99.99/mo and 20x at $199.99/mo — adding Deep Think reasoning, the highest Deep Research access, video generation headroom, and 20–30TB storage. It's aimed at daily heavy users of premium features; most people saturate AI Pro first.
No — the allowance refreshes every 5 hours, inside a weekly ceiling, matching the structure Anthropic uses for Claude. Heavier features drain the pool faster, so there's no fixed number of prompts per window.
Providers change these limits without notice, and several are deliberately unpublished compute budgets rather than fixed counts. Everything above was checked against the linked sources on September 2, 2026; if you spot something outdated, email hello@userightai.com.
Newsletter
Providers adjust caps quietly. We re-verify and note every change, dated.
No spam. Useful updates only. Affiliate disclosures always clearly labeled.