Anthropic’s surprisingly powerful new model, Fable, is worth trying out while you can. Built by the company behind chatbot Claude, Fable is the publicly secure version of Mythos, the model Anthropic deemed too dangerous to launch just two months ago. To make it available to the average user, Anthropic has introduced strict barriers designed to prevent the cutting-edge model from being used for high-risk applications.
Those who have used it describe Fable as a step-change in AI capability, capable of handling complicated tasks with ease. That’s one reason to adopt it… and quickly. But there’s another: in less than two weeks, it will disappear from most people’s budgets.
For now, Anthropic monthly subscribers can access Fable with Pro, Max, Team, or Enterprise plans. But that’s only until June 22. After that, Anthropic will put it solely behind API access, where users have to pay per token. And if there’s one thing that big businesses and everyday users are quickly realizing as their bank accounts are hit by ever-increasing AI bills, it’s that tokens aren’t free.
Anthropic says limited subscriber access is due to capacity limitations. “As there is enough capacity online, our goal is to make it a standard part of those [subscription] limits again,” Anthropic’s head of growth, Amol Avasare, wrote about X. “We are running as hard as we can on this,” he added. But Avasare said the company couldn’t make any promises about the timing.
However, what seems like a temporary capacity problem increasingly feels like an indication of the future direction of AI. Since the launch of ChatGPT in November 2022, consumer AI has been built on a simple premise: pay $20 or more a month, get access to the world’s smartest machines, and use them as much as businesses allow. Most users never reached their limits, while many regular users got effectively subsidized access.
But as agent AI replaces chatbots as the newest way to interact with AI systems, the price of inference (the cost of providing the service) has skyrocketed. The better these systems become, the more likely people will be to ask them to do longer, more complicated, and more valuable work, even as the cost of the hardware needed to service the models continues to rise.
That’s why the industry is quietly moving from “eat all you can” to “eat what you can afford.” Last month, Google moved Gemini toward compute-based limits; Microsoft’s GitHub recently reduced its usage limits; And in late May, OpenAI disabled its usage limit multiplier after a month-long test.
All are examples of the never-ending AI subscription model being squeezed by the economic realities of delivering the technology. The era of “unlimited” frontier AI was fun while it lasted.