Google is making a seriously calculated entrance into September. The company has launched its latest AI models, Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, while Alphabet simultaneously received a major legal reprieve after a U.S. federal judge rejected the Justice Department’s demand that Google sell its AdX advertising exchange.

The timing could hardly be more important. Alphabet has just endured a four-month stock-market losing streak, its longest monthly decline since 2015, leaving investors hungry for proof that its huge AI spending can eventually translate into sustainable growth. Now Google has another AI model to sell, its advertising infrastructure remains largely intact, and Wall Street has a fresh reason to watch what happens next.

More Trending Stories: A routine unpaid bill in Bangkok spiraled into something far darker. For more, read our story on the host bar bald-shaving incident under police investigation.

Gemini 3.8 Flash Takes Google’s AI Strategy Into Coding and Autonomous Agents

Google’s latest offensive is centred on Gemini 3.8 Flash, its third Flash model released in only six weeks.

Google describes Gemini 3.8 Flash as its most capable reasoning and coding model to date. The company says it brings significant improvements over Gemini 3.7 Flash in software engineering, agentic tasks and difficult multi-step reasoning, while maintaining the same introductory pricing.

The model is available at $0.75 per million input tokens and $3.75 per million output tokens. That introductory price will remain in place through December 31, 2026. Beginning January 1, 2027, Google says pricing will increase to $1.50 per million input tokens and $7.50 per million output tokens.

Official detail	Gemini 3.8 Flash
Model	Gemini 3.8 Flash
Focus	Coding, reasoning, agentic workflows
Input pricing	$0.75 per million tokens
Output pricing	$3.75 per million tokens
Introductory pricing ends	December 31, 2026
New pricing from	January 1, 2027
Future input price	$1.50 per million tokens
Future output price	$7.50 per million tokens
Release pace	Third Flash model in six weeks

The bigger story, however, isn’t simply another version number.

Google is increasingly targeting the rapidly expanding market for AI agents and enterprise software development. Instead of AI systems that merely answer prompts, businesses are looking for models capable of handling longer sequences of tasks, using tools and completing complex work with less human intervention.

Google says Gemini 3.8 Flash can approach the performance of larger frontier models on certain tasks while operating at substantially lower cost. Its own testing places the model ahead of several larger models on the DeepSWE v1.1 long-horizon software-engineering benchmark.

That cost equation matters enormously for enterprise customers. When an AI application processes millions or even billions of tokens, the price of inference can determine whether a product makes commercial sense.

Tulsee Doshi, Google’s senior director of product management, summed up the company’s reaction to the rapid progress of its Flash models:

“really surprised us in positive ways in their performance,”

Doshi added:

“give us opportunities to lean into them.”

The strategy reflects a wider shift across the AI industry. The race isn’t only about building the biggest possible frontier model anymore. Smaller, faster and cheaper systems capable of delivering strong results are becoming increasingly important, particularly when businesses need AI running continuously.

More Odd-Theft News: Nearly half a million oysters disappeared overnight in a heist that’s left one farmer counting years of losses. For more, read our post on the A$350,000 NSW oyster heist.

Gemini 3.8 Flash Cyber Brings Google AI Into High-Stakes Cybersecurity

Google wasn’t finished with one model.

Alongside Gemini 3.8 Flash, the company introduced Gemini 3.8 Flash Cyber, specifically designed for cybersecurity applications.

Google describes it as its most capable cybersecurity model, with frontier-level performance in vulnerability detection and automated patching. The model is being introduced through Google’s new Fairwind Program, with initial access restricted to trusted government authorities, critical infrastructure operators and software maintainers.

That restricted rollout isn’t accidental.

Cybersecurity AI is inherently dual-use. A system that can discover and repair vulnerabilities could become an extremely powerful tool for defenders, but similar capabilities could also potentially be abused by attackers.

Google says Gemini 3.8 Flash includes safeguards designed to address misuse involving cyber offense as well as chemical, biological, radiological and nuclear (CBRN) risks. The company has also highlighted a significant improvement in protection against prompt-injection attacks.

Doshi said:

“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.”

Google’s own announcement says Gemini 3.8 Flash Cyber is being made available to trusted defenders through Fairwind, while the general Gemini 3.8 Flash is available to developers, enterprise users and Google AI Pro and Ultra subscribers.

More Strange Sightings: A floating landmass appeared out of nowhere on a Canadian reservoir — then vanished just as mysteriously. For more, read our story on the disappearing island at Williston Reservoir.

Google Is Making AI Pricing a Major Part of Its Enterprise Strategy

There’s another piece of the Gemini 3.8 Flash strategy that could prove just as important as raw benchmark performance: price.

Google is keeping Gemini 3.8 Flash at its predecessor’s introductory token rates despite the performance improvements the company says it has achieved.

The company is also changing the economics surrounding Gemini Enterprise, introducing pay-as-you-go access, token discounts of up to 20%, monthly limits on agent spending and a zero-dollar base subscription option.

Google is positioning those options against enterprise AI products from Microsoft and Anthropic. Its argument is straightforward: competing offerings can become more expensive when businesses have to pay recurring seat fees alongside separate product licences.

That could become increasingly relevant as companies move from experimenting with AI to deploying agents at scale.

And Google has something that many newer AI competitors simply cannot reproduce overnight: an enormous existing cloud and enterprise customer base.

Google Cloud CEO Thomas Kurian told CNBC that nearly three-quarters of Google Cloud customers are already using Google’s AI products. Those customers, he said, are spending roughly 50% more than their original commitments.

That creates a potentially powerful route for Alphabet to monetise AI. Google doesn’t necessarily have to build an entirely new customer ecosystem because it can sell AI into relationships it already has.

More Real-Life Drama: A smart-home gadget meant to clean floors ended up exposing a marriage — and landing someone behind bars. For more, read our story on the robot vacuum affair that sent a husband to prison.

DeepMind’s AI Agent Strategy Could Be Bigger Than Gemini 3.8 Flash

The latest Gemini launch also comes during a period of major change inside Google’s AI organisation.

DeepMind CEO Demis Hassabis has described a future in which Gemini could operate as a general-purpose layer capable of coordinating cheaper specialised models and AI agents.

Hassabis discussed that strategy at the G20 Innovation meeting on Wednesday, marking his first major public appearance since DeepMind underwent a restructuring last month. As part of that restructuring, Hassabis moved from CEO of the unit to chairman.

The direction is revealing.

Google’s strategy increasingly suggests that it doesn’t believe one gigantic model needs to win every single task. Instead, the company could combine powerful general-purpose models with smaller specialised systems designed to handle particular jobs more efficiently.

That could mean Gemini serves as the central intelligence while specialised models and agents do the heavy lifting underneath.

For Google, that vision fits neatly with its existing advantages in cloud infrastructure, search, advertising, mobile software and enterprise services.

The company already owns much of the infrastructure needed to distribute AI at enormous scale.

More Viral Buzz: A 47-year-old biohacker’s quest for immortality just took a detour through the dunk contest. For more, read our post on Bryan Johnson’s 35-inch vertical dunk challenge.

Is Google Still Behind OpenAI and Anthropic in Enterprise AI?

Here’s where the excitement around Gemini 3.8 Flash meets a much tougher reality.

Google’s latest model may strengthen its position, but not everyone believes it changes the overall enterprise AI hierarchy.

Gil Luria, an analyst at D.A. Davidson, told CNBC that Gemini 3.8 Flash keeps Google competitive but doesn’t fundamentally eliminate the company’s gap with leading enterprise AI providers.

Luria said:

“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.”

That is the uncomfortable question hanging over Alphabet.

Google has extraordinary technical resources, a massive distribution network and huge computing capacity. Yet OpenAI and Anthropic have established powerful positions in enterprise AI and developer workflows.

So the challenge isn’t simply proving that Google can make impressive AI.

It has to prove that people and businesses will actually use it at scale.

The real test is whether Gemini can generate meaningful revenue growth, increase Google Cloud spending, strengthen enterprise adoption and protect the company’s existing businesses.

More Workplace Shockers: A missed sales target ended with a hospital visit after an extreme physical punishment went too far. For more, read our story on the employee hospitalized with rhabdomyolysis after a 200-squat punishment.

Google’s Advertising Business Gets a Major Legal Reprieve

While Gemini is grabbing the technology headlines, Google’s other big September development happened inside a courtroom.

Google’s advertising business remains one of Alphabet’s biggest financial engines. The company’s advertising business grew 14% in the latest quarter, generating the cash Alphabet needs to fund its enormous AI infrastructure spending.

Then came the AdX ruling.

U.S. District Judge Leonie Brinkema rejected the Justice Department’s request to force Google to sell its AdX advertising exchange.

But calling the decision a complete Google victory would be misleading.

Brinkema had previously ruled that Google unlawfully monopolised the publisher ad-server and ad-exchange markets and unlawfully tied its DFP and AdX products together.

Wednesday’s decision was about the remedy, not whether Google violated antitrust law.

Instead of ordering the structural breakup requested by the government, Brinkema opted for behavioral remedies.

The detailed ruling was temporarily sealed while the parties reviewed confidential information and proposed redactions, meaning the complete scope of the restrictions wasn’t immediately public.

More Bizarre Hoaxes: A “creepy sighting” that spread panic across two countries turned out to be nothing but pixels. For more, read our story on the AI-generated Cat in the Hat hoax police were forced to debunk.

Why the DOJ Wanted Google to Sell AdX

The Justice Department and a coalition of states argued that Google should divest AdX, the advertising exchange that connects buyers and publishers through real-time auctions.

The government also sought additional structural measures involving Google’s publisher advertising technology, including changes concerning DFP’s auction technology.

The distinction is important.

Google avoided the structural breakup the DOJ wanted, but the underlying antitrust finding remains.

The government’s original 2023 case alleged that Google used its position across the advertising technology stack to weaken competitors and preserve its dominance.

In April 2025, Brinkema found Google liable for monopolising the publisher ad-server and ad-exchange markets. The court concluded that Google’s conduct harmed publishers, competition and consumers.

The DOJ subsequently pushed for stronger structural remedies, including the sale of AdX.

But Google argued that separating AdX from its broader advertising technology infrastructure would be technically difficult and disruptive for publishers, advertisers and other customers.

The judge ultimately agreed that a forced divestiture wasn’t necessary.

More Fandom Events: Wizarding-world fans are getting a nighttime hunt through the Dark Arts this Halloween season. For more, read our story on Tokyo’s first-ever Dark Arts Halloween at the Warner Bros. Studio Tour.

Google Celebrates, But the DOJ Says the Fight Isn’t Over

Following the ruling, Lee-Anne Mulholland, Google’s vice president of regulatory affairs, said:

“We’re very pleased the Court rejected the DOJ’s proposal to break apart tools that help small businesses reach new customers and grow.”

The Justice Department took a different tone while acknowledging that the court had imposed restrictions.

The department said:

“The Antitrust Division is pleased that the court ordered substantial relief in the Google Ad Tech case,”

It added:

“We are one step closer to restoring competition and bringing relief for the American people in online advertising markets. The Department is evaluating appropriate next steps.”

That tells you almost everything about why the ruling is complicated.

Google gets to keep AdX.

The DOJ still gets court-ordered remedies.

And the larger antitrust fight isn’t simply disappearing.

More Trending Stories: What started as a dispute over an unpaid bill escalated into an alleged public humiliation now under police scrutiny. For more, read our article on the Bangkok host bar coercion investigation.

Alphabet Keeps Its Advertising Infrastructure Intact—For Now

For Alphabet, the practical significance is huge.

AdX remains inside Google’s advertising ecosystem. The company therefore avoids the major structural separation the Justice Department had sought.

At the same time, Google will have to comply with behavioural restrictions ordered by the court.

And those details matter.

Because the full remedies order was temporarily sealed, it would be premature to describe the decision as an unrestricted Google victory or suggest that its advertising practices are no longer under meaningful scrutiny.

The most precise takeaway is this:

Google avoided the forced sale of AdX while remaining subject to behavioral remedies following the court’s earlier monopoly finding.

That distinction could become increasingly important as the full details of the remedies emerge.

Berkshire Hathaway Adds Another Vote of Confidence in Google’s AI Strategy

Alphabet also received a notable vote of confidence Wednesday from Berkshire Hathaway CEO Greg Abel.

Abel told CNBC that Berkshire views Alphabet as a winner in AI, partly because Berkshire’s portfolio companies provide an inside look at how Google’s technology is being used.

Abel said:

“We have a lot of visibility from within our companies as to how we’re using AI, what type of benefits it’s delivering, so that brought incremental interest, and then we saw Google as a significant player,”

The timing matters.

Investors are trying to determine whether Alphabet’s enormous AI investments will eventually produce stronger earnings and market share.

Google is spending heavily on servers, data centres and networking infrastructure to support its AI ambitions.

The numbers show just how enormous that commitment has become.

Alphabet reported $27.9 billion in capital expenditures during the fourth quarter of 2025 and $91.4 billion for the full year, with the majority directed toward technical infrastructure.

That is an extraordinary amount of money.

And it means Alphabet needs its AI businesses to eventually generate substantial returns.

More Odd-Theft News: Investigators are still untangling how thieves managed to pull off Australia’s biggest oyster heist. For more, read our post on the 400,000-oyster NSW theft and its long recovery.

Google Stock Still Has Something to Prove After a Brutal Four-Month Slide

Despite all the positive developments, investors haven’t completely forgotten the company’s recent problems.

Alphabet entered September following a four-month losing streak, its longest monthly decline since 2015.

The stock fell more than 1% at the start of September before recovering somewhat Wednesday.

Alphabet shares ultimately rose approximately 0.6% Wednesday, but that wasn’t enough to erase the recent weakness.

That leaves Google at a critical point.

The company is releasing Gemini models at an increasingly rapid pace. It has a new cybersecurity offering. It has a potentially powerful enterprise pricing strategy. Its cloud customers are already using Google’s AI products.

And now it has avoided the immediate threat of an AdX sale.

But Wall Street still needs evidence.

The question is whether Gemini’s rapid development can translate into real enterprise adoption, higher cloud spending and stronger monetisation.

At the same time, Alphabet has to keep its advertising machine running while adapting to the restrictions imposed by the federal court.

More Strange Sightings: An entire floating mass drifted onto a reservoir, sparked theories for weeks, then quietly disappeared. For more, read our story on the vanishing island mystery at Williston Reservoir.

Google’s September Comeback Has a Lot Riding on Gemini and AdX

Put everything together and Alphabet’s opening to September suddenly looks much more interesting.

Gemini 3.8 Flash strengthens Google’s push into coding, reasoning and autonomous AI agents. Gemini 3.8 Flash Cyber takes that technology into cybersecurity, with access initially limited to trusted defenders.

Google’s pricing strategy is designed to appeal to businesses increasingly focused on AI inference costs, while Gemini Enterprise gets new pay-as-you-go options and token discounts of up to 20%.

The AdX ruling simultaneously removes the immediate threat of a forced sale of one of Google’s most important advertising technology assets.

But this isn’t a clean victory.

Google still has to compete against OpenAI and Anthropic in enterprise AI. Its infrastructure spending remains enormous. Its stock has just come through its longest monthly losing streak in more than a decade.

And, critically, the antitrust fight hasn’t vanished.

The AdX decision preserves Google’s ownership of the exchange, but it does not overturn the court’s finding that Google unlawfully monopolised parts of the digital advertising market.

That leaves Alphabet with an unusually high-stakes September.

It has the technology.

It has the infrastructure.

It has the customers.

It still has the advertising machine.

Now Google needs to prove that all of those advantages can come together before investors lose patience.

For Alphabet, the story isn’t really about one new Gemini model or one antitrust ruling. It’s about whether the company can turn its enormous technological and financial resources into sustained AI growth without breaking the advertising engine that helped finance the entire race.

For now, Google has another Gemini model to sell, another major antitrust threat partially behind it and another chance to convince Wall Street that its AI strategy can actually deliver.

More Real-Life Drama: A privacy case built on footage from a household cleaning robot ended up deciding a marriage’s fate. For more, read our story on the robot vacuum evidence that led to a prison sentence.

Disclaimer

This article is based on information available as of September 3, 2026, and has been prepared using a thorough review of the sources identified below, including Google’s official Gemini announcement and model documentation, U.S. Department of Justice materials concerning the Google ad-tech antitrust case, Alphabet investor-relations information, CNBC reporting and other published reports covering the Gemini launches, Alphabet’s stock performance and the AdX remedies ruling. Details of Google’s court-ordered advertising remedies may change or become clearer when the court releases the fuller redacted opinion. Statements regarding Gemini 3.8 Flash and Gemini 3.8 Flash Cyber’s performance reflect claims or assessments by Google and identified third parties and should not be interpreted as independent guarantees of performance.

Sources

• Google — Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
• U.S. Department of Justice — Google Ad Tech Antitrust Case
• U.S. Department of Justice — April 2025 Google Digital Advertising Antitrust Ruling
• U.S. Department of Justice — Google Ad Tech Case Documents and Remedies
• Alphabet Investor Relations — 2025 Fourth-Quarter and Full-Year Results
• Alphabet Investor Relations — 2025 Q4 Earnings Call
• CNBC — Google starts September with AI momentum after longest monthly losing streak in over a decade
• Associated Press — Judge orders changes to Google’s digital ads business but spares it from a breakup
• Ars Technica — US court rules Google will not have to sell ad exchange

Featured Image Credit: Matheus Bertelli on Pexels