AGI Pulse
ProductGoogle · Google

Google limits Gemini app models by plan from October 9: free accounts keep only Flash-Lite, AI Plus loses Pro

0 reads

Published
2026-10-08
Type
Product
Organization
Google
Source
Google

Why it matters

【Key points】

Google's Gemini Apps help page says model availability for personal accounts changes from October 2026: the changes "start to take effect for users without an AI subscription on October 9th", while AI Plus subscribers "should receive an email that explains when these changes will take effect for you".

The new availability table: no plan, Flash-Lite only; AI Plus, Flash-Lite and Flash but no Pro; AI Pro and AI Ultra, all three.

Each available model also gains an effort level, "low, medium, and high", which the page says improves answers but "will also use more of your limit".

【Verifiable】

Primary: https://support.google.com/gemini/answer/17004136 .

The same page describes compute-based usage limits that "refresh every 5 hours until you reach your weekly limit"; media generation, Deep Research, the Pro model and Deep Think burn them faster. The limit table reads: no plan, standard limits; AI Plus 2x and AI Pro 4x standard; AI Ultra "5x or 20x higher than AI Pro depending on your subscription".

One sentence is unstable. The English page now reads "If you have an AI Ultra subscription you’ll also have access to the Deep Think option for maximum parallel reasoning." The October 4 (UTC) archived copy of the same English page, and the Simplified Chinese version still live today, both say AI Pro or AI Ultra. The table matches in all three.

【Our read】

Cutting the free tier to Flash-Lite changes what free is for: not a taste of the flagship but a funnel, with inference cost per free user down and the gap a subscription buys visible on first use.

Effort levels put a price on thinking: the same model costs more of your allowance when it reasons harder.

AI Plus losing Pro, with no firm date and notice by email, suggests the middle tier is still being tuned; plan differences are shifting from which models you get to multiples of a quota.

Until Google's own pages agree, treat Deep Think on AI Pro as unsettled: the policy is still moving.

Read original →

Related developments

Aggregated by AGI Pulse. Titles and links only; no full text is republished. Editorial policy