On this page
Anyone searching Google's support pages in late 2026 for a fixed daily prompt cap—such as 50 messages per day—will no longer find one. On May 17, 2026, Google transitioned consumer Gemini Apps to compute-based metering, replacing static prompt counters with a dynamic allowance governed by two simultaneous clocks: a 5-hour rolling refresh window and a 7-day weekly ceiling.
Every tier from Free to the top $199.99 Ultra plan operates under this dual-clock structure. Subscribers can inspect live usage percentages and exact reset timestamps at any time on gemini.google.com by opening Settings in the lower-left sidebar and selecting Usage limits. Unused allowance does not roll over between windows.
The Five Consumer Tiers and Per-Unit Cost Math
Google structures personal Gemini access across five tiers: Free, Google AI Plus, Google AI Pro, Google AI Ultra (5x), and Google AI Ultra (20x). Because Google expresses paid allowances as multipliers of the Free baseline rather than fixed message counts, dividing each monthly price by its multiplier reveals the underlying unit economics:
**Free Tier ($0 / month): **Establishes the 1x baseline allowance with a 32K-token context window (roughly 48 pages of text) and access to Flash-Lite alongside capped image generation, Deep Research, Gemini Live, and Canvas. Once the 5-hour allowance is exhausted, Free users wait until the next 5-hour refresh, and compute-heavy tools like Deep Research are throttled on Free first during peak demand.
**Google AI Plus ($4.99 / month in the US, or $9.99 / month with 2 TB storage): **Doubles the Free allowance to 2x, expands the context window to 128K tokens (about 190 pages), unlocks the full Flash model and video generation, and includes 200 monthly Google Flow AI credits plus 400 GB of cloud storage. At $4.99 for 2x capacity, Plus works out to roughly $2.50 per 1x unit—the lowest entry price per compute unit for light text users who do not need a 1M context window.
**Google AI Pro ($19.99 / month or $199.99 / year): **Delivers 4x the Free allowance and a full 1M-token context window (up to roughly 1,500 pages of text), unlocking Gemini 4 Argon, Gemini 4 Hydrogen, and full Pro models. Bundled perks include 1,000 monthly Flow credits, 200 Colab compute units, $10/month in Google Cloud credits, higher limits in Gemini Notebook (renamed from NotebookLM on July 16, 2026), entry-level paid agent quota in Google Antigravity, and 5 TB of storage. At $19.99 for 4x capacity, Pro sits at roughly $5.00 per 1x unit.
**Google AI Ultra 5x Tier ($99.99 / month, added May 19, 2026): **Scales capacity to 5x the Pro allowance (20x Free) with a 1M context window, unlocking Gemini 4 Carbon and Deep Think reasoning, 10,000 monthly Flow credits, 2,000 Colab compute units, $40/month in Google Cloud credits, and 20 TB of storage. Its per-unit cost remains around $5.00 per 1x unit, scaling linearly for heavy daily workflows.
**Google AI Ultra 20x Tier ($199.99 / month, reduced from $249.99): **Provides 20x the Pro allowance (80x Free), top scheduling priority and concurrency across Gemini and Google Antigravity, 25,000 monthly Flow credits, and 30 TB of storage. Because capacity quadruples over the 5x tier while the price doubles, the effective unit cost drops back to around $2.50 per 1x unit.

Image source: compiled by DigIntoAI from Google One AI Plans and Google Help documentation.
Four Factors That Drive Compute Burn Rate
Under compute-based metering, two ten-prompt sessions can consume vastly different shares of your 5-hour and weekly meters depending on four variables:
**Model and Thinking Depth: **Running Gemini 4 Carbon or Gemini 4 Argon—especially with high reasoning effort or Deep Think enabled—consumes substantially more compute per turn than Gemini 4 Hydrogen or Flash-Lite.
**Compounding Thread Length: **Every follow-up prompt inside an active thread re-processes the accumulated conversation history. Once a chat grows to hundreds of thousands of tokens, even a short follow-up prompt consumes several times more weekly allowance than starting fresh. A single 50-turn thread drains far more weekly quota than five separate 10-turn chats.
**Tool and Media Invocations: **Launching Deep Research reports, generating video clips, or running multi-step code execution consumes large discrete blocks of your 5-hour window.
**File Upload Size and Duration Caps: **Google allows up to 10 files per prompt at up to 100 MB per non-video file (2 GB max per video file). Cumulative video duration caps at 5 minutes on Free versus 1 hour on Pro and Ultra; cumulative audio caps at 10 minutes on Free versus 3 hours on Pro and Ultra; and code folders or GitHub repositories cap at 5,000 files and 100 MB total (ZIP archives cannot contain audio or video).

Image source: compiled by DigIntoAI from Google Help and Google Antigravity documentation.
Fallback Behavior, Practical Habits & What This Is Not
Several operational boundaries distinguish how Gemini handles quota exhaustion and cross-product access:
**Automatic Fallback to Flash-Lite Rather Than a Hard Lockout: **When paid subscribers (Plus, Pro, or Ultra) hit their 5-hour or weekly high-tier ceiling, Gemini does not terminate the conversation; it automatically falls back to the lightweight Flash-Lite model until the relevant clock resets. Ultra plans do not include an unlimited slow queue—they operate as a fixed high-multiplier compute pool, though Pro and Ultra subscribers can purchase optional AI Credit top-ups for Flow or Antigravity overages.
**Match the Model to the Task and Reset Threads Frequently: **Default to Gemini 4 Hydrogen for drafting, formatting, and routine research, reserving Argon or Carbon for dense technical synthesis. Opening a New Chat as soon as a topic concludes prevents stale context from compounding your weekly burn.
**Promotional Trials and Regional Bundles in Developer Clients: **First-month discounted Pro promotions or carrier-bundled Pro subscriptions receive the full 4x web/app multiplier and 1M context window inside gemini.google.com. However, developer clients such as Google Antigravity perform separate checks on full-price billing status and regional eligibility, meaning some trial or regional promo accounts remain restricted to Antigravity's free starter quota.
**Workspace and Enterprise Plans Follow Separate Rules: **The 2x, 4x, 20x, and 80x multipliers apply exclusively to personal Google One AI subscriptions. Commercial Google Workspace tiers (Business Starter, Standard, and Plus at $7, $14, and $22 per user/month) and Gemini Enterprise (from $21/seat/month) rely on organizational pooling or fixed daily feature caps—such as 20 versus 100 images per day, or Education limits of 25 Pro prompts per 4 hours and 5 Deep Research reports per month.
