Paying subscribers on the $4.99 monthly plan lose Pro model access while free app users are restricted to the Flash-Lite model under Google’s Gemini AI access changes beginning October 9, 2026.
Personal accounts accessing the Gemini app without a paid subscription will be limited to just the 3.5 Flash-Lite model, removing default access to the fuller Flash and Pro variations that were previously available. The rules governing how everyday users interact with Google’s artificial intelligence assistant are undergoing a major restructuring.
The adjustments touch both non-paying users and those on the entry-level paid tier. According to support documentation published ahead of the effective date, the change applies only to users accessing Gemini via the web and app using personal accounts, leaving work and school accounts outside the immediate scope of the downgrade. Subscribers to AI Plus will receive an email from Google outlining the exact date the transition affects their specific account.
Google Reorganizes Gemini Model Tiers
Under the revised structure, the tier boundaries determine exactly which intelligence engines answer a user’s prompt. Free users lose access to both the 3.6 Flash and 3.1 Pro models, leaving them with Flash-Lite as their sole option for day-to-day queries.
The cheapest paid subscription takes a hit as well. Subscribers to the Google AI Plus tier, which costs $4.99 a month in the US (or £4.49 / AU$7.99) and ₹399 per month in India, will retain Flash-Lite and Flash models but will no longer have access to the Pro model.
At $4.99 per month, AI Plus members now find themselves on a subscription tier that closely resembles the complimentary option, augmented only by the inclusion of Flash alongside Flash-Lite. What they will miss is Pro’s ability to work through harder, longer tasks with more care.
Users who rely on advanced reasoning for programming or complex analysis face a stark choice: maintain the previous capability ceiling by upgrading to higher tiers or accept the constraints of smaller models. Higher-end plans remain fully stocked.
| Subscription Plan | Flash-Lite | Flash | Pro |
|---|---|---|---|
| Free Tier | Yes | No (After Oct 9) | No (After Oct 9) |
| AI Plus ($4.99/mo) | Yes | Yes | No (After Oct 9) |
| AI Pro ($19.99/mo) | Yes | Yes | Yes |
| AI Ultra | Yes | Yes | Yes |
Google Extends Reasoning Modes to AI Pro Subscribers
To sweeten the deal for upgrades, Google is extending advanced features downward to the AI Pro subscriber base. AI Pro users will now get Deep Think, a reasoning mode previously reserved for the priciest plans.
“maximum parallel reasoning.”
Google, via Notebookcheck
Alongside model restrictions, the platform is rolling out new effort settings across models this month. Users can select between Low, Medium, and High effort levels to balance response depth against resource consumption.
Raising the effort level helps the assistant work through complex tasks and deliver thorough answers, but it will use more of a user’s usage limit. Usage quotas continue to refresh every five hours until weekly caps are reached.
Computing Costs Drive Model Restrictions
The tightening of model access reflects rising operational expenses across the artificial intelligence sector. Running frontier models incurs enormous computational costs, a financial reality highlighted when OpenAI posted a $38.5 billion net loss despite generating $13.07 billion in revenue in 2025.
Industry competitors have made parallel moves. OpenAI halved advanced-model allowances on select paid plans, while Anthropic restricted Claude Opus 5.5 to paying customers, demonstrating an industry-wide transition from free access to capture users toward tiered operations to boost paid conversion.
While consumer tiers face these restrictions, Google has also introduced a specialized model named Gemini 4 Argon. Argon is initially available only to cybersecurity professionals and U.S. government agencies through the Fairwind program due to safety considerations surrounding its advanced software vulnerability capabilities.
Google chief AI architect Koray Kavukcuoglu addressed the cautious deployment strategy in a company blog post, stating that Safely releasing frontier capabilities at this level requires a phased approach.
Which specific work and school accounts will eventually be included in these restrictions remains unstated by Google?