Gemini usage limits: the five-hour window and the weekly limit

Short answer: Gemini meters use by compute rather than a fixed number of prompts. Your limit refreshes every five hours until you reach a weekly limit, and long chats, heavy models and features use it faster. Paid Google AI plans raise the limits.

Gemini used to be described in prompts per day. It no longer works that way, which is why old figures you find online do not match what you see. Two limits now decide when Gemini stops, and the length of your conversation is one of the things that spends them.

How the limits work

• The five-hour window. Your allowance refreshes every five hours. Run through it and Gemini stops until the window comes round.
• The weekly limit. The five-hour refreshes continue only until you reach the week's total. After that, waiting a few hours does not help.

To see what is left, open Settings at the bottom left of gemini.google.com and select Usage Limits. Gemini also warns you as you near a limit, and says when it refreshes once you reach it.

Neither is a count of messages. Google says the limits factor in the complexity of your prompt, the models and features you use, and the length of your chat. A short question to a fast model costs little; a long answer from the most capable model, late in a long conversation, costs much more.

Why it ran out after a few prompts

A frequent complaint in forums since the change: the limit arrived after three or four prompts. Usually the prompts were not ordinary chat. Google rates what each kind of request costs:

• High: Deep Research and video generation.
• Medium: ongoing conversations and large file analysis.
• Very low: a simple conversation in chat.

So a handful of research reports or videos can use a five-hour window that hundreds of short questions would not. With a Google AI subscription, you can keep the conversation going with Flash-Lite after reaching the limit.

Free and paid plans

Google AI plans for personal accounts, such as Google AI Plus, Pro and Ultra, give expanded access to models and features and higher limits. Work and school accounts are managed differently, by the organisation. Plans and the models each includes change often, so check Google's own plan page rather than a figure in an article.

Why long chats use it faster

The model does not keep a conversation in memory between messages. Each new message sends the conversation back to be read again, so message two hundred carries the other hundred and ninety-nine with it. That is why chat length is one of the factors Google names. How the context window works.

How to make it last

1. Start a new chat when the topic changes. Carry the decisions across, not the transcript. A handover that works.
2. Match the model to the task. Save the most capable model for the work that needs it.
3. Ask for the change, not the whole thing. Only the edited part of a long document costs less than a full rewrite every time.
4. Front-load heavy work. Large files and long analysis early in a window; quick questions near the end of it.

How Tvara shows it

Tvara reads the five-hour window, the weekly limit and the AI credit balance from Gemini's own figures and puts them where you are working, with reset times. The one that will stop you soonest comes first, and Tvara warns you at 20% and again at 10% left. It keeps separate figures for each Google account you are signed in with, because each has its own allowance.

It only reads what Gemini already reports. Nothing is estimated and no request is changed. Allowance tracking is free. Tvara on Gemini · Claude's limits · ChatGPT and the rest