Google has detailed exactly how many prompts can be used in a 24-hour period with its Gemini chatbot's free, AI Pro and AI Ultra tiers. Previously, Google was more vague on how many questions were permitted under each tier.
Per Google's Gemini support page, those using the AI chatbot for free have up to five prompts on its 2.5 Pro model a day. For Gemini AI Pro, a subscription tier that costs $20 per month, you'll be allowed 100 prompts. With Gemini AI Ultra at $200 a month, that number is 500 prompts. When it comes to the context window (or the amount of text measured in tokens that the model can consider at one time to generate a response), free users get 32,000 tokens, whereas both AI Pro and AI Ultra users get 1 million tokens.
In Deep Think mode, in which Google's AI models can perform more in-depth analysis, free members are allowed five reports a day with the lighter 2.5 Flash model. Subscribers paying for AI Pro can call up to 20 reports, and those with AI Ultra can call 200 reports, both using the 2.5 Pro model.
Don't miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source.
Image generation is limited to 100 images per day on the free tier. Under AI Pro and AI Ultra, subscribers can create 1,000 images per day. Google says Nano Banana is the default model for image generation across Gemini.
Video generation using Google's Veo 3 model is unavailable to those on the free tier. With AI Pro, Google gives you three videos per day, whereas with AI Ultra, you'll get five per day. All tiers are allowed up to 20 audio overviews a day.
Google said it didn't have an additional comment when asked.
AI companies tend to keep exact usage rates and token limits vague. For example, in OpenAI's FAQ on ChatGPT's free tier, it says that ChatGPT will notify you when you've hit your usage for GPT-4o within a 5-hour window, but doesn't specify how much. (GPT-4o has since been retired in favor of GPT-5, though GPT-4o is still available for Plus users.)
Until it updated its support page, Google kept information about usage limits unclear, likely because not all queries require the same compute power. Also, AI models are constantly evolving, either becoming more efficient or requiring greater power for stronger reasoning capabilities. Given how rapidly the space is changing, it was probably easier for companies to avoid having to update usage and token rates on a regular basis.