Google just slashed the price of its new Gemini 4 Neon model by 25% below its flagship Carbon tier — and the establishment press is calling it a win for consumers when it's really a war chest maneuver designed to starve out startups and open-source competitors before they can mount a challenge.
The stakes are straightforward: whoever controls the AI infrastructure layer controls the next economy. When a trillion-dollar company uses its ad-revenue hoard to price below what any scrappy competitor can match, it doesn't stay cheap forever. The price goes back up once the competition is dead.
Here's what happened. Google launched Gemini 4 Neon, a mid-range AI model that matches or exceeds its heavier Carbon model on key benchmarks at three times the speed, according to Thai Tran, Glean's AI product lead, quoted in Google's announcement. Neon costs $3 per million input tokens and $15 per million output tokens. That's 25% below Carbon's scheduled standard rates of $4 and $20, which kick in after promotional pricing expires December 1.
VentureBeat framed the price cut as a "welcome surprise" and praised Neon as "affordable, too, especially compared to similarly scoring proprietary models." 9to5Google kept it clinical — a product rollout post with benchmark tables. Neither outlet raised the obvious question: how long can a startup or an open-source project survive when Google prices a competitive model a quarter below its own flagship rates?
Neon replaces Argon in Google's model lineup and slots between Hydrogen, the lightweight workhorse, and Carbon, the heavy-duty model. It ranks third overall on the Artificial Analysis Intelligence Index and first on both the Legal Research Bench and Tax Agent Bench, according to Google's own figures. It also tops Zapier's AutomationBench and ranks second on the new Agent Arena Index. By every public measure, this is a serious model priced aggressively against Anthropic's Opus 5.5 — which costs $4 and $20 per million tokens — while beating it on select engineering benchmarks.
The pricing structure tells the real story. Google is running promotional rates on Carbon and Argon through November — $2 input, $10 output — before the hike. That means developers building on those models today will face a 100% output-price increase in weeks, right as Neon arrives as the "affordable" alternative. It's a classic bait-and-switch funnel: hook developers on cheap rates, herd them onto the new model, then squeeze once the ecosystem is locked in.
Both outlets reported Neon's headline rates matter-of-factly. VentureBeat noted that under OpenAI's long-context pricing, a GPT-6.2 Sol request exceeding 272,000 input tokens costs $4 input and $15 output — effectively matching Neon. That's the kind of competitive pressure that forces a price war only incumbents with cash reserves can survive.
The press celebrates lower prices. Fair enough — cheaper API access helps developers today. But the same playbook ran in oil, railroads, and telecom: price the competition into the ground, consolidate, then extract. When three companies control the foundation models and can cross-subsidize from other business lines, the word "competition" stops meaning what the founders intended when they broke monopolies at the tavern.
The open question is whether antitrust enforcers will notice before the AI market belongs entirely to those who already own everything else.








