If we’re having trouble keeping up with all the AI news this week, we can’t imagine how you feel.
The last three days have been a nonstop parade of new products, features, and frontier model releases from the leading AI labs. There are new names like “Sol” and “Terra,” as well as new features for image creation and voice assistants. Microsoft added a splash of weird with its “poetry pharmacy.”
There were also major personnel changes. First, OpenAI’s chief futurist left after nine years. Then OpenAI’s CEO of appplications, Fidji Simo, stepped down after her medical leave.
Behind the scenes, Perplexity is quietly working on a new coding tool called “Teammate” to compete with Anthropic and OpenAI, Business Insider exclusively reported on.
Even investor Jason Calacanis can’t keep up. “It’s like a better new operating system, laptop and CPU being launched every 14 days,” he wrote on X.
We know the AI labs like to one-up each other, but this week underscored that you shouldn’t expect a summer slowdown in the AI wars. Here’s a rundown of all the important updates:
OpenAI had the most announcements
On Wednesday, the Sam Altman-led lab debuted GPT-Live, a new voice model meant to make ChatGPT Voice feel more like talking to a human. Yes, that means you can talk over it.
The next day, OpenAI debuted the GPT-5.6 family of models. This included the flagship frontier model Sol, the everyday Terra, and the cost-saving Luna. It also integrated Codex into the ChatGPT desktop app and announced a new workplace-oriented agent called ChatGPT Work.
As the company cuts down on its “side quests,” OpenAI also announced that it was sunsetting its Atlas browser.
Meta eyes the enterprise AI market
After breaking its silence with Muse Spark in April, Meta’s superintelligence unit led by Alexandr Wang has released a string of products.
The company debuted a new image-generation model on Tuesday, called Muse Image. The product sparked controversy after public Instagram accounts were opted in by default.
Next, Meta released the next iteration of its Muse model, Muse Spark 1.1. It’s the first model Meta has charged for, and CEO Mark Zuckerberg said it had a “very low price.”
Zuckerberg also called out the other AI labs for pricing their chatbots at “very extreme” levels.
SpaceXAI and Anthropic — friends at last?
Once rivals, the newly rebranded SpaceXAI and Anthropic now seem to be getting along just fine. Elon Musk said he was “clearly wrong about Anthropic” and that it was the clear AI leader (Anthropic has also been paying SpaceX $1.25 billion a month for compute).
But beyond the sentiment shift by Musk, his AI lab also company released its latest model, Grok 4.5, on Wednesday. Musk described it as an “Opus-class model.” It’s also the first model the company has released since its acquisition of Cursor.
On Thursday, Anthropic released a new “Reflect” feature, where users could see a summary and analysis of their habits with Claude.
