Google clarifies Gemini AI Plus limits as NotebookLM lands inside the iOS app

Google Gemini

• Google launches Gemini AI Plus in US at $7.99/month with expanded prompt limits
• AI Plus offers higher caps than free tier but fewer than Pro and Ultra plans
• Gemini app on iOS now integrates NotebookLM for enhanced content creation

Did our AI summary help?

Google has officially rolled out its Gemini AI Plus plan in the US and, alongside the launch, has published a clearer breakdown of how the subscription changes usage limits inside the Gemini app. The new tier sits between the free experience and the more expensive Pro and Ultra plans, offering higher caps but stopping well short of Google’s top-end AI allowances.

Under AI Plus, which costs $7.99 per month, users move beyond the free “Basic access” tier and gain expanded prompt limits across multiple Gemini models. With the Thinking model, powered by Gemini 3 Flash, subscribers can send up to 90 prompts per day. That is a noticeable step up from the free experience, but still significantly lower than the 300 daily prompts available on the AI Pro plan.

The more capable Gemini 3 Pro model also sees tighter restrictions on AI Plus. Subscribers are limited to 30 prompts per day, compared to 100 prompts on the $19.99 Pro plan. This positions AI Plus as a mid-tier option aimed at regular users rather than heavy, all-day AI workloads.

Context size is another area where Google draws clear lines between plans. Free Gemini users are capped at a 32,000-token context window, which roughly translates to about 50 pages of text. AI Plus expands that to 128,000 tokens, allowing users to work with much larger documents or longer conversations without losing context. For comparison, AI Pro and Ultra push this much further, offering up to a one million token context window, or around 1,500 pages of text.

Not every feature sees an upgrade with AI Plus. Image generation limits remain unchanged, with Nano Banana capped at 100 images per day and Nano Banana Pro limited to three images daily, matching the free plan. AI Plus subscribers do, however, get limited access to Veo 3.1 Fast, Google’s video-focused AI tool, though the company has not specified exactly how restrictive that access is.

Deep Research access also remains constrained. Google lists this feature as “limited access using Pro”, suggesting that AI Plus users do not receive a meaningful upgrade here. Scheduled actions are similarly unchanged, with users still able to run up to 10 at a time, the same limit applied to free accounts.

Google has also confirmed that these usage limits apply equally to subscribers on the Google One 2 TB plan, which bundles Gemini benefits with cloud storage. In other words, paying for storage does not unlock higher AI caps beyond what AI Plus already provides.

Alongside these pricing clarifications, Google has quietly expanded Gemini’s functionality on iOS. This week, the Gemini app on iPhone gained built-in access to NotebookLM through the plus menu. The integration preserves NotebookLM’s native design while allowing users to directly pull notebooks into Gemini prompts.

Once connected, users can create new content based on a notebook, add a notebook to a custom Gem, or apply other Gemini tools such as Canvas, Veo, Guided Learning, or Deep Research using the notebook’s contents as context. The same integration is also rolling out to Google Workspace users, extending the feature beyond personal accounts.

According to Google, bringing NotebookLM directly into Gemini is meant to help users build on deep, structured knowledge bases rather than starting from scratch with each prompt. By grounding conversations in existing notebooks, Gemini can generate more relevant responses and more accurately tailored content.

Together, these updates show Google refining its Gemini monetisation strategy. AI Plus offers tangible improvements over the free tier, but it is clearly designed to funnel power users toward higher-priced plans. At the same time, deeper integrations like NotebookLM suggest Google is betting on ecosystem stickiness rather than raw prompt volume alone.