Google's latest mid-size model, Gemini 3.8 Flash, is set at $0.75 for a million input tokens and $3.75 for a million output tokens, aligning with the prior Flash's introductory rates. The pitch is that the compact Flash variants cost less to operate, are simpler to iterate on, and on certain workloads are narrowing the gap with its biggest frontier systems. This release calls out better performance across coding, agent workflows, and reasoning.
According to Google, Gemini 3.8 Flash Cyber identifies and fixes software vulnerabilities with frontier-level results, while running much faster and at lower cost than larger models. Given the potential for misuse, early access will be restricted to a select set of vetted government and enterprise defenders via the new Fairwind Program. "We're really excited about being able to provide an offering to defenders that is a fraction of the cost, much faster, while still showcasing that frontier-level performance," Doshi said.
Gemini Enterprise is introducing pay-as-you-go pricing, token discounts reaching up to 20%, monthly caps on what agents can spend, and a base subscription set at $0. In materials provided to CNBC, Google contrasted itself with Microsoft and Anthropic, arguing that ongoing seat charges and distinct product licenses push up costs and curb flexibility. Scale is part of the case as well: almost three-quarters of Google Cloud customers already use its AI, and CEO Thomas Kurian told CNBC that the same customers now spend roughly 50% above their original commitments.
DeepMind's Demis Hassabis sketched a longer-term view at the G20 Innovation meeting, saying Gemini may evolve into a broad orchestration layer that coordinates lower-cost specialized models and agents. It was his first public remarks since last month's reorganization, in which he moved from DeepMind CEO to chairman of the unit.
The money story behind the headlines is usually the part that matters. Join Market Briefs free and get it every morning.
Price matters, but the leaderboard is still the leaderboard. "From a product perspective this model seems to keep Google in the race, but probably won't change the fact they are a distant third in the enterprise market," said D.A. Davidson's Gil Luria, who recommends holding the stock. At the same time, Berkshire's Greg Abel told CNBC's Becky Quick that Berkshire views Alphabet as one of the AI winners, pointing to how its portfolio companies are using Google's tech.
The ad business continues to hum, up 14% in the latest quarter. And a federal judge in the Justice Department's ad-tech case chose behavioral remedies and concluded that Google does not need to divest its AdX exchange, following a separate ruling last year rejecting calls to force a Chrome divestiture. For your wallet, the through-line is simple to watch: whether aggressive pricing plus Google's scale can convert those AI investments into durable customer spend.
