Google changed which Gemini models each plan can reach, and the change cut deepest at the free end. This page is the plan-by-plan answer, kept in one table.
Checked on 9 October 2026 against Google's Gemini model access support page and 9to5Google. Google revises these limits often, so treat the support page as the final word.
The table
Model access by plan, from 9 October 2026:
| Plan | Flash-Lite | Flash | Pro | Deep Think |
|---|---|---|---|---|
| Free | yes | no | no | no |
| AI Plus (about $4.99 a month) | yes | yes | no | no |
| AI Pro | yes | yes | yes | no |
| AI Ultra | yes | yes | yes | yes |
What changed on 9 October: free users lost Flash and Pro, and now reach Gemini 3.5 Flash-Lite only. AI Plus subscribers kept Flash-Lite and Flash but lost Pro. AI Pro and AI Ultra saw no change in model access.
What each model tier is for
The tier name tells you the trade-off, and it is the same shape at every AI company:
| Tier | What it is for |
|---|---|
| Pro | the most capable: complex reasoning, long documents, hard coding |
| Flash | the speed and value workhorse: high-volume, real-time, everyday tasks |
| Flash-Lite | the cheapest: simple tasks called in very large numbers |
| Deep Think | an extended reasoning mode, on the highest consumer plan only |
So a free user has not lost Gemini. They have been moved to the tier built for simple, high-volume work, which is a real step down for study and research tasks that benefit from a stronger model.
The new effort settings
The Gemini app also added low, medium and high effort settings. Higher effort means the model reasons for longer before answering, which produces better answers on hard problems and uses more of your usage limit. On a constrained plan, effort is now a budget you spend, not just a quality dial.
How people reacted
The change was not well received. Users raised concerns on Google's Gemini community forum, calling the restrictions unfair and saying they limit what can be done for study and research.
If you need a stronger model free
The consumer app is not the only door. The Gemini API has a free tier covering Flash-class models, with lower rate limits than the paid tier. It is one of the most generous ways for a student to experiment with a frontier-style model.
The trade-off is a real one: on the free API tier Google may use your data to improve its products, while the paid tier does not carry that term. So use it to learn and build, not for anything private.
Full detail on the four ways into Google's stack is in Gemini, Antigravity and Project Astra, and the difference between paying by plan and paying per token is in API vs subscription.
Common questions
Which Gemini model do free users get?
From 9 October 2026, free users of the Gemini app get Gemini 3.5 Flash-Lite only. Access to Gemini Flash and Gemini Pro was removed for the free tier.
Does AI Plus still include Gemini Pro?
No. From 9 October 2026 AI Plus subscribers, who pay about 4.99 dollars a month, keep Flash-Lite and Flash but lose access to Pro. To reach Pro you need AI Pro or AI Ultra.
Did AI Pro and AI Ultra change?
No. Model access for AI Pro and AI Ultra users stayed as it was before 9 October 2026. Deep Think remains exclusive to AI Ultra.
Can I still use Gemini Pro for free in any way?
Not in the Gemini app. The Gemini API has a free tier, but it covers Flash-class models rather than Pro, and it comes with lower rate limits and a term allowing Google to use free-tier data to improve its products.
What do the low, medium and high effort settings do?
They control how long Gemini reasons before answering. Higher effort gives better answers on hard problems and consumes more of your plan's usage limit, so on a smaller plan it is worth saving high effort for questions that need it.
Links
- Google's model access support page: support.google.com/gemini/answer/17004136
- Google AI plans: one.google.com/about/google-ai-plans
- Gemini API pricing: ai.google.dev/gemini-api/docs/pricing
- 9to5Google on the change: 9to5google.com