Google plans to change model availability for Gemini Apps personal accounts in October 2026, with changes for users without an AI subscription scheduled to start Oct. 9. Under the reported plan, the models available would vary by subscription tier.
Which models are planned for each tier?
The reported allocation distinguishes free accounts from Google AI Plus, Pro and Ultra:
| Personal-account tier | Reported model access |
| Free | Gemini Flash-Lite 3.5 only |
| Google AI Plus | Gemini Flash-Lite 3.5 and Gemini Flash 3.6; no Pro |
| Google AI Pro | Gemini Pro 3.1 |
| Google AI Ultra | Gemini Pro 3.1 |
Deep Think and effort settings
Deep Think is reported to become available to Google AI Pro subscribers as well as Google AI Ultra subscribers. The option had previously been limited to Ultra.
Gemini’s reported task-effort settings are low, medium and high. Choosing higher effort can use more of the processing and compute limit, so a task’s settings affect how much of that allowance it consumes.
How Gemini usage limits work
Google reportedly shifted Gemini usage limits in May 2026 from a fixed message count to limits based on processing capacity and actual compute use. The Gemini API has separate rate limits, measured in requests per minute, tokens per minute and requests per day.
