Google trims Gemini model access for free and AI Plus users from October 9
Google will restrict free and AI Plus Gemini users to fewer models from October 9, while AI Pro gains a Deep Think mode.
Under the new arrangement, users who do not hold a Google AI subscription will be limited to the Flash-Lite (3.5) model, which Google describes as offering the fastest answers. The Flash (3.6) model for all-around help and the Pro (3.1) model for advanced reasoning will no longer be available to them. Subscribers to AI Plus, priced at $4.99, £4.49 or AU$7.99 a month, lose access to Pro (3.1) but keep Flash (3.6).
The AI Pro and AI Ultra tiers do not lose any models. AI Pro gains a Deep Think mode that maximizes parallel reasoning at the cost of some speed. Google said affected users will be notified through a message inside the Gemini app or by email, and that AI Plus subscribers will receive an email setting out the timetable for the change.
Until now, Google's consumer Gemini plans differed chiefly through usage limits: users could reach the same models but were given different numbers of queries. The update adds model availability to the differences between tiers. Google says exact limits depend on the complexity of a prompt, the features used and the length of a chat, and that paid users have higher limits than users without a Google AI subscription. Limits refresh every five hours, and weekly caps also apply.
The change brings Gemini closer to the way OpenAI structures ChatGPT. OpenAI also gates model access by monthly payment, though it applies separate limits to different models, query types and thinking modes rather than a single allowance.
TechRadar noted that AI computing capacity does not come cheap, and said Google had presumably calculated that the new restrictions would save it money and might persuade more users to sign up for a subscription.