Google Increases Gemini 3 Usage Limits With New Version Split

Google has changed how users access its Gemini 3 models, moving away from a shared usage system to independent limits for each version. This update came after users complained that they couldn’t properly plan which model to use for their work.

When Gemini 3 first launched in December, both the Thinking and Pro models pulled from the same pool of available prompts. If you were paying for the AI Pro plan, you got 100 prompts per day total for both models. With AI Ultra, that number went up to 500 prompts per day, but again, you had to share that limit between both versions.

This approach created confusion for people who needed to switch between different types of work. Someone might burn through their daily limit using the Thinking model for problem solving, only to find they had nothing left when they needed the Pro model for coding later in the day.

Google responded by separating the two models entirely. Now, AI Pro subscribers get 300 prompts per day for the Thinking model and another 100 prompts per day for the Pro model. These limits work independently, meaning using one doesn’t affect the other.

For AI Ultra subscribers, the numbers jump significantly. They now have access to 1,500 Thinking prompts per day and 500 Pro prompts per day, all running on separate counters.

The Thinking model is designed to handle difficult problems quickly. It uses Gemini 3 Flash and focuses on solving challenges that need reasoning but don’t require deep specialization. Meanwhile, the Pro model runs on Gemini 3 Pro itself and is built for advanced mathematics and programming work where accuracy matters more than speed.

Free users also benefit from the change, though Google hasn’t published exact numbers for their tier. The company only mentions that free accounts get “Basic access” to both models, and these limits can change daily depending on how many people are using the service.

The timing matters because Google has been dealing with high demand for Gemini 3 since its release. In late November, the company reduced free access to its image generation tool, Nano Banana Pro, to just two images per day because servers were getting overwhelmed.

This latest update shows Google is trying to give users more control and transparency about how they use its services. People who rely on AI for different types of work throughout the day can now plan their usage better without worrying about running out of access to the wrong model at the wrong time.

The Fast model remains available with broader access for everyday tasks and quick queries. Google positions it as the option for situations where you need speed but don’t require the advanced capabilities of Thinking or Pro.