Google just pulled the classic tech company move: get users hooked on free models, then yank the ladder up once they're dependent.
The Summary
- Starting October 9th, free Gemini users lose access to Flash and Pro models, getting locked into Flash Lite only
- The $4.99/month AI Plus tier also gets gutted, losing Pro model access entirely
- Pro and the new "Deep Think" reasoning feature now require Google AI Pro or Ultra subscriptions (pricing unannounced)
- This is Google admitting inference costs still matter, even as competitors race to zero
The Signal
Google's pricing squeeze comes right as the AI market is splitting into two camps: those betting inference gets cheap enough to give away, and those who think premium reasoning is the moat. Google is now firmly in camp two, despite spending the last year acting like camp one.
The timing matters. OpenAI just launched o1-preview with extended thinking time. Anthropic is pushing Claude's analysis capabilities. Google responds by walling off its Pro model and introducing "Deep Think," a feature that sounds suspiciously like reasoning you have to pay for. The free tier becomes a demo, not a product.
"Free users can currently choose from Gemini Flash Lite, Flash, and Pro, but now you'll need a $4.99/month Google AI Plus subscription to access the standard Flash model."
Here's what the three-tier squeeze looks like:
- Free: Flash Lite only (basic queries, limited context)
- $4.99 AI Plus: Flash access, but no Pro
- AI Pro/Ultra (price TBA): Pro model plus Deep Think reasoning
The middle tier is particularly telling. Google is creating a paid tier that will immediately feel inadequate. You'll pay $5/month and still hit walls when you need deeper analysis. That's not a product strategy. That's a funnel to the higher tiers.
Google says AI Plus subscribers will get emails about when they lose Pro access, which means this is a staggered rollout. They're testing how much pushback they get before pulling the trigger completely. Classic enterprise move: announce the bad news, gauge the reaction, adjust if the outcry is too loud.
The Implication
If you're building anything that depends on free or cheap AI inference, this is your canary. Google has more money to subsidize free usage than almost anyone, and they just decided it's not worth it. Expect OpenAI, Anthropic, and others to follow with their own tier compression. The race to the bottom on pricing is over. The race to justify premium pricing has begun.
For individuals: if you're using Gemini casually, Flash Lite might be fine. But if you've built it into any workflow, budget for the upgrade now or start testing alternatives. For companies: the bet that AI gets cheap enough to treat as a commodity just got riskier.