Google just dropped three new Gemini AI models with slashed pricing, using rock-bottom costs to lock developers into its censorship-prone infrastructure while quietly building a government-exclusive cybersecurity tool the public can't touch.
The move is a classic monopolist play—undercut on price, embed your product everywhere, and make it too expensive for free-speech alternatives like X's Grok or independent platforms to compete. With Google's well-documented history of baking progressive orthodoxy into Gemini's outputs, cheaper doesn't mean freer. It means the censorship gets more efficient.
Google released Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber on Tuesday. The headline number: 3.6 Flash uses up to 17% fewer tokens than its predecessor, and output pricing dropped from $9 to $7.50 per million tokens, according to Ars Technica. Flash-Lite goes even lower—$0.30 per million input tokens and $2.50 per million output tokens—making it Google's cheapest modern AI. That's the kind of pricing designed to starve out competitors who can't run at a loss backed by Alphabet's cash reserves.
The performance, though, tells a different story. Wccftech reports that Gemini 3.6 Flash scored just 50 on the Artificial Analysis Intelligence Index—behind Grok 4.5, GPT-5.6 Luna, GPT-5.6 Terra, Claude Sonnet 5, and Meta Spark 1.1. Google is selling a below-average model at below-market prices to buy market share. Benzinga notes Alphabet stock still carries a premium valuation at 26.8x earnings, with analysts projecting $113.6 billion in revenue this quarter—meaning Google can afford to bleed on token prices until the competition folds.
Then there's Gemini 3.5 Flash Cyber, a model fine-tuned for finding and fixing cybersecurity vulnerabilities. Google says it will be available exclusively to governments and "trusted partners" through a limited-access pilot. Ars Technica reports that Google, borrowing a page from Anthropic, frames the model as too dangerous for public release due to its "dual-use" nature. A powerful cyber tool that only the state and its handpicked allies can access—what could go wrong?
Meanwhile, the flagship model Google actually promised—Gemini 3.5 Pro—is nowhere to be found. TechCrunch reports that Google teased Pro back in May, saying it was "already being used internally" and would roll out the next month. June came and went. Bloomberg reported internal delays as the model failed to meet performance goals. Google is now rushing cheap models out the door while its top-tier product stalls.
Ars Technica also notes that Google has already deprecated Gemini 3.5 Flash—released just two months ago at I/O—and is replacing it with 3.6 Flash in the Gemini app and API. That aggressive churn forces developers onto the latest censored version whether they want it or not. And Flash-Lite is being baked directly into Google Search's AI Overviews, meaning the censored outputs reach hundreds of millions of users who never opted in.
Google's strategy is clear: flood the zone with cheap, compliant AI, lock in developers with low prices, gatekeep the dangerous tools for the state, and let the flagship model wait until it meets internal standards. The question is whether the market rewards a mediocre, subsidized product just because it's cheap—or whether users and developers will choose performance and freedom over a discount.
