Google Launches Gemini 3.8 Flash, Wins Bid to Keep $0.75-Per-Million Ad Exchange

3 articles · Updated · CNBC · Sep 3

Google paired its Gemini 3.8 Flash rollout with a court win Wednesday after a federal judge rejected the Justice Department’s bid to force a sale of AdX, choosing behavioral remedies over a breakup.
Gemini 3.8 Flash is Google’s third Flash model in six weeks, priced at $0.75 per million input tokens and $3.75 for output, with a new Cyber version initially limited to trusted government and enterprise defenders.
The launch leans on lower-cost, faster-to-iterate models as Google tries to rebuild AI momentum after DeepMind’s restructuring and talent losses, though one analyst still called it a distant third in enterprise AI.
Google is also widening its enterprise pitch with pay-as-you-go Gemini pricing, discounts of up to 20% and spending caps, while saying nearly three-quarters of Google Cloud customers already use its AI products.
For Alphabet investors, the combined AI and legal developments offer relief after a difficult summer: the stock rose 0.6% Wednesday, Berkshire’s Greg Abel backed Google as an AI winner, and courts have now twice declined to order major divestitures.

Sources

Left40%

Center60%

CNBC8d ago

Google Launches New AI Models, Wins Ad Exchange Antitrust Case

Ars Technica8d ago

Google Releases Gemini 3.8 Flash AI Model, Its Best Reasoning and Coding Offering

The Keyword | Google Product and Technology News8d ago

Google Introduces Gemini 3.8 Flash and 3.8 Flash Cyber AI Models

10 Sources

As Google gates its powerful cyber AI behind closed doors, are layered defenses truly enough to stop inevitable prompt injection breaches?

If Google's new AI agent autonomously hunts and patches vulnerabilities, what happens when hackers reverse-engineer its defensive logic?

With AI now running endless agentic loops for complex coding, will the hidden token costs silently bankrupt enterprise developers?

Gemini 3.8 Flash and Flash Cyber Launch: Google Escalates the 2026 AI Price-Performance War with Breakthrough Coding and Cybersecurity Models

Overview

In August and September 2026, Google accelerated the AI competition by launching three Gemini Flash models in just six weeks, culminating with Gemini 3.8 Flash and its cybersecurity-focused variant. This rapid release cycle intensified the price and performance war with OpenAI and Anthropic, forcing enterprises to move away from slow procurement and adopt continuous evaluation strategies. Gemini 3.8 Flash delivers advanced reasoning and tool use, but at higher operational costs, so Google still supports the more cost-efficient Gemini 3.7 Flash. For cybersecurity, Gemini 3.8 Flash Cyber is strictly gated to vetted defenders, balancing powerful vulnerability detection with strong safety controls to prevent misuse.

...