Google Launches Gemini 3.8 Flash With Same Pricing as Predecessor, Its Third Flash Model in Six Weeks

Google released Gemini 3.8 Flash on Wednesday, its third Flash-tier model in six weeks, pricing it at $0.75 per million input tokens and $3.75 per million output tokens — identical to the introductory pricing of Gemini 3.7 Flash. Google calls it its best reasoning and coding model yet, claiming significant improvements over 3.7 Flash in software engineering and multi-step agentic tasks.

Alongside the general-purpose release, Google launched Gemini 3.8 Flash Cyber, a cybersecurity-focused variant the company says can detect and patch software vulnerabilities at "frontier-level performance" while running faster and cheaper than larger models, according to Google. Access to the cyber model is initially restricted to a small group of government and enterprise customers through a new program called Fairwind, reflecting concerns about dual-use misuse.

Tulsee Doshi, senior director of product management at Google DeepMind, told CNBC the recent Flash models have "really surprised us in positive ways in their performance," giving the company room to expand their role instead of relying solely on larger frontier systems. Smaller Flash models cost less to run and iterate faster than Google's largest models, while narrowing the performance gap on select tasks.

Gil Luria, an analyst at D.A. Davidson, was more measured, telling CNBC the release "seems to keep Google in the race, but probably won't change the fact they are a distant third in the enterprise market" behind Anthropic and OpenAI. Luria currently rates the stock a hold.

Pricing pressure on rivals

Google is pairing the release with pricing changes to Gemini Enterprise, adding pay-as-you-go billing, token discounts of up to 20%, monthly caps on agent spending, and a zero-dollar base subscription tier. According to materials Google provided to CNBC, the company is directly targeting Microsoft and Anthropic, arguing their recurring seat fees and separate product licenses make their offerings costlier and less flexible than Google's.

Google Cloud CEO Thomas Kurian told CNBC nearly three-quarters of Google Cloud customers already use its AI products, and that those customers are spending roughly 50% more than their original commitments — a scale argument Google is leaning on as it competes for enterprise share.

Context for the launch

The release lands as Alphabet closes its longest monthly losing streak on Wall Street since 2015, a four-month stretch marked by high-profile departures and a DeepMind reorganization that saw Demis Hassabis move from CEO to chairman of the unit. Hassabis, speaking publicly Wednesday for the first time since that reorganization, told the G20 Innovation meeting that Gemini could increasingly function as a coordination layer for cheaper, specialized models and agents — a strategy where breadth matters as much as raw model quality.

The same day, a federal judge rejected the Justice Department's push to force Google to divest its AdX ad exchange, opting for behavioral remedies instead of a structural breakup. Berkshire Hathaway CEO Greg Abel also told CNBC his firm views Alphabet as an AI winner, citing visibility into how Berkshire's portfolio companies use Google's technology.

Alphabet shares rose 0.6% Wednesday after falling more than 1% the prior day, leaving the stock still slightly down for the young month.

What this means: Gemini 3.8 Flash is not a frontier-model challenge to GPT-5-class or Claude Opus-class systems — it's a mid-tier, cost-optimized release aimed at defending Google's enterprise pricing position while Anthropic and OpenAI continue to lead flagship benchmarks. The unchanged pricing versus 3.7 Flash signals Google is competing on cost-per-task rather than raw capability gains, betting that cheap, fast iteration and cloud-scale distribution matter more than topping leaderboards. The antitrust win and Abel's endorsement are separate wins for Alphabet's stock story, but neither changes the competitive reality Luria describes: Google remains a distant third in enterprise AI despite shipping models at a faster cadence than its two closest rivals.