… Update 5/19/26 : A day after this story was initially published, Google announced more changes to its AI plans, including a new $100-a-month AI Ultra plan, while the top-tier AI Ultra subscription is now $200 a month, a $50/month price cut. …
… But with the rise of agentic AI functionality such as vibe-coding applications and “computer use” abilities, flat-rate AI subscribers are burning far more tokens than ever before, and the big AI providers are struggling to keep up with the demand. …
… Then there’s Haiku 4.5, a lightweight model that’s ideal for straightforward tasks like proofreading, summarizing news articles, and making grocery lists based on recipe ideas. …
… Of course, the big AI providers are responding in kind, frantically adjusting and readjusting the usage limits on their flat-rate plans–sometimes invisibly–as hoards of everyday AI users unleash armies of AI agents on their desktops. …
… Most sophisticated models can't fit inside the limitations of your PC's memory and storage. Phison and Intel are working on a solution. Using Phison's aiDAPTIV solution, a 26 billion parameter AI model can be run on a laptop with 16GB of RAM, versus the 32GB of RAM it would normally require. …
… During weekdays between 5am–11am PT / 1pm–7pm GMT, you'll move through your 5-hour session limits faster than before." And in a comment to our colleagues at Infoworld, analyst Pareekh Jain of Pareekh Consulting explained further: "The impact is largely limited to individual users, prosumers, and sm… …
… The other factor to consider is the jaw-droppingly fast pace of AI development, meaning that today’s hot-rod AI hardware might not be a hot-rod in, say, two years. For its part, AMD is pitching its AI Developer Platform as a way for the Ryzen AI Halo to keep pace in the months and years ahead. …
… From faster search to better virtual desktops and even AI-powered tools like Copilot, the OS is geared toward helping you get things done in fewer steps. …
… Anthropic—who also struck computing capacity deals with Amazon, Google, and Microsoft—said it was also interested in SpaceX's orbital AI compute capacity. …
… With a custom mobile processor optimized for local AI, on-device agents could handle more mundane duties like triaging your email or suggesting the best nearby lunch spots, perfect for keeping agentic costs down while ideally maintaining your privacy. …