Google dropped the most powerful IDE

Gemini 3 turns coding into orchestration

A new model rarely changes how a keyboard sounds. This one might.

Google’s Gemini 3 lands as the company’s most capable system yet, built for reasoning-heavy work and priced at $2 per million input tokens in preview. It slots straight into the tools developers already touch every day, from Android Studio to GitHub plug-ins, and scores 54.2 percent on Terminal-Bench 2.0, a test that measures how well a model can operate a real terminal.

The shift is clearest in Antigravity, Google’s new development environment where parallel agents draft features, refactor files, and wander through browsers while you act more like a project architect than a coder. It feels oddly physical, like watching multiple cursors crawl across a dim monitor as the fan hums. This is Google’s answer to long-horizon coding work, a bet that developers want orchestration more than autocomplete.

Pricing, context window, and tool depth signal an arms race for the IDE itself, not just the models that feed it. When multimodal agents can read a screen, trace a gesture, and adjust a codebase in one sweep, the boundary between “idea” and “app” gets thin.

And somewhere in that narrowing gap sits the next fight over who owns the workflow.

OpenAI just wired Intuit into your chat window

Soon, your tax refund starts in a chat box.

OpenAI and Intuit signed a multi-year, $100M-plus deal that puts TurboTax, QuickBooks, Credit Karma, and Mailchimp directly inside ChatGPT while Intuit doubles down on OpenAI’s models across its own products. For OpenAI, Intuit joins more than 1 million business customers already building on its stack, but this one sits right on your money.

This is about pipes. The partnership turns ChatGPT into a front door for concrete financial actions, from estimating a refund to nudging payroll to forecasting cash, with Intuit’s proprietary data and credit models flowing quietly behind the scenes while you stare at a glowing laptop at 11:47 p.m. in April.

The money is loud. If this works, every bank, broker, and payroll provider has to decide whether they become an app tile inside ChatGPT or race to build a rival assistant that can match this kind of direct account access.

The real question is who your finances will feel like they belong to once the chat window becomes the control panel.

Chinese researchers still anchor America’s A.I. push

Seven of Zuckerberg’s 11 star hires came from China.

The new studies from Carnegie and alphaXiv land with a thud because they show how deeply U.S. labs depend on Chinese researchers who arrived years before the boom and stayed through the visa snarls and political noise. They also show that collaboration between the two countries has remained the most common pairing in global A.I. papers since 2018.

Some analysts worry that Washington’s pressure campaign could throttle this pipeline at the exact moment when companies like Meta and Microsoft are pouring billions into systems that require an enormous bench of senior researchers whose training spans multiple institutions and two languages. A crowded conference hallway in San Francisco tells the story more clearly than any speech because Mandarin drifts through the air beside the clatter of laptops.

The numbers add weight, especially the 87 out of 100 top-tier Chinese researchers who stayed in the United States after 2019. This stickiness signals that the country still offers the fastest path to scale, credit, and compute.

A shift in visa policy could change that overnight.

Still, talent moves where it feels welcome, and that part of the race remains unwritten.

Prohuman team

Free Guides

Explore our free guides and products to get into AI and master it.

All of them are free to access and would stay free for you.

Feeling generous?

You know someone who loves breakthroughs as much as you do.

Share The Prohuman it’s how smart people stay one update ahead.