+ SpaceX's AI Group Released A Model That Undercuts The Leaders On Price
On July 8, SpaceXAI launched Grok 4.5, the first model it built together with Cursor since buying the coding company for 60 billion dollars. On the standard industry tests, it performs in the same range as the top models from Anthropic and OpenAI, while running faster and using far less computing to do the same work. SpaceXAI says it is roughly four times more efficient than Anthropic's flagship.
The number that matters is the price. Grok 4.5 costs less than half of what Anthropic charges for its top model, and it is free for now inside Cursor and SpaceXAI's own tools. Elon Musk called it an Opus class model that is faster, cheaper, and lighter to run, and said a bigger one is coming next month. It is not yet available in Europe, which is expected in mid July.
What it all means: A year ago Grok was treated as an also-ran. It just matched the best models in the world at a fraction of the price, and it did it on efficiency rather than raw power. The frontier is getting crowded at the top and cheap at the same time, and any AI budget set this spring is already too high.
+ Microsoft Cut 4,800 Jobs And Spent 2.5 Billion Dollars To Help Companies Use AI
On July 6, Microsoft laid off about 4,800 people, roughly 2 percent of its staff, with its Xbox and sales teams hit hardest. Around 1,600 of the cuts were in Xbox, with more expected later this year. Microsoft's own people chief said the eliminated roles are not being replaced by AI, while acknowledging that AI is changing how the work gets done. It was part of a wider stretch that has seen about 154,000 tech jobs cut in the first half of 2026.
Days earlier, Microsoft launched Microsoft Frontier Co., a 2.5 billion dollar business of about 6,000 people whose job is to sit inside customer companies and build AI tools for them on-site. Amazon announced a one billion dollar version of the same idea two days before that. Both companies are betting on the same problem: plenty of firms now have AI and cannot figure out how to use it, and closing that gap is worth billions.
What it all means: The money is moving from building models to putting them to work. Microsoft and Amazon spent the week betting that the bottleneck is adoption, and the layoffs are the other half of that bet. Routine roles are shrinking while the work of installing AI inside a real business is where the budgets now go.
+ Meta Built Its Own AI Image Maker Instead Of Renting One
On July 7, Meta released Muse Image, the first picture-making model from the new Superintelligence Labs group led by Alexandr Wang. It opened at number two on the public leaderboards for creating and editing images, behind only OpenAI. Unlike older tools that turn a prompt straight into a picture, Muse Image plans the layout first, can blend several photos, pulls in live information from the web, and edits its own results to improve them.
It is free inside the Meta AI app and is rolling out across Instagram and WhatsApp, with Facebook and Messenger to follow. Meta also teased a video version coming soon. The bigger shift is behind the scenes: Meta used to pay outside companies to power its creative tools, and it just replaced them with something it owns.
What it all means: For a company whose products run on photos and video, owning the image engine outright matters as much as any feature it ships. Meta just cut out the outside suppliers it depended on and put a free, capable image maker in front of the billions of people already inside its apps. Everyone who paints, designs, or sells with images now has one more strong free option, built by the company that owns their feed.
