For a long time, the deal with AI model developers was straightforward: enjoy free access to an online chatbot or subscribe monthly for more usage, premium features, and advanced models. This simple bargain is about to become much more intricate, thanks to a significant change from Anthropic.
Effective July 12 at 11:59 PM PT, subscribers to Anthropic’s existing $20, $100, and $200-a-month plans will face additional usage-based fees. These new charges will apply specifically to access Claude Fable 5, the consumer version of their highly capable Mythos 5 AI model. This move marks what appears to be the first instance of a frontier AI lab introducing usage-based billing for a consumer AI model.
A New Chapter for AI Pricing
The new rates for Claude Fable 5 will mirror those already in place for developers using Anthropic’s API. Users will be charged $10 for every million tokens sent to Claude and $50 for every million tokens the model generates in response to their queries. To put this into perspective, if a $20/month subscriber sends one million tokens and receives one million tokens in a month, they would owe an extra $60, bringing their total monthly bill to $80.
While a million tokens may sound like an astronomical amount—roughly equivalent to 750,000 words, which is longer than the entire Lord of The Rings book series—it’s not uncommon for AI power users to accumulate substantial API bills. This is partly because newer, more sophisticated AI models like Fable 5 often utilize many more tokens in a hidden “chain-of-thought” process to formulate comprehensive answers. Historically, AI labs have favored flat monthly subscriptions for consumers, using them to generate predictable revenue and manage demand, contrasting with the “pay-as-you-go” standard for developers.
Why the Shift? Unpacking the Reasons
This shift towards usage-based billing isn’t entirely out of the blue; the AI industry has been signaling this direction for a while. Last year, AI coding startups like Cursor revised their unlimited AI subscriptions in favor of usage-based pricing models. More recently, Anthropic itself began charging its large business clients based on their employees’ AI usage rather than a flat fee. These changes could also be part of a broader strategy to streamline Anthropic’s financials ahead of a potential initial public offering.
Some prominent AI executives argue that traditional subscription plans are becoming obsolete, especially with the rise of AI agents like Claude Code and Codex. These advanced agents demand significantly more computational power than conventional chatbots. Nick Turley, formerly OpenAI’s head of ChatGPT and now leading the company’s enterprise products, eloquently put it earlier this year: “It’s possible that, in the current era, having an unlimited [AI] plan is like having an unlimited electricity plan. It just doesn’t make sense.”
Anthropic, however, hasn’t completely closed the door on all-inclusive subscriptions. According to spokesperson Reem Ateyeh, the company aims to reintegrate Fable 5 into Claude’s standard subscription plans “when sufficient capacity allows,” and intends to do so “as quickly as we can.” This statement hints at the company’s current computational constraints, despite Anthropic securing multi-billion dollar deals for data center capacity with major players like SpaceX, Amazon, and Google. Yet, it remains uncertain when, if ever, Anthropic will overcome these capacity limitations to fully restore Fable 5 to its subscription offerings.
Testing Consumer Appetite and Market Position
This pricing adjustment follows an extended promotional period during which Fable 5 was offered to subscribers at no extra cost. Anthropic initially anticipated “very high, and difficult to predict” demand for the model, a forecast that proved accurate. Interest in Fable 5 surged further after a brief, high-profile ban by the US government for foreign nationals, followed by its subsequent general approval on July 1. This new billing structure is undoubtedly a critical test of consumer willingness to pay for Anthropic’s AI models, particularly as the company increasingly ventures into the consumer space, historically dominated by OpenAI’s ChatGPT and Google’s Gemini.
Claude has been experiencing rapid growth, reaching 245 million unique visitors in May—more than double its February figures, according to market intelligence firm Sensor Tower. While this is still a considerable distance from ChatGPT’s 1.11 billion and Google Gemini’s 662 million monthly unique visitors, Claude’s upward trajectory is undeniable. This rising popularity coincides with some criticism from technology leaders who accuse Anthropic of charging premium prices while leveraging vast amounts of intellectual property.
In essence, Anthropic is betting on becoming the “Apple of the AI Era,” confident that a segment of users will always be prepared to pay a premium for what they perceive as the best AI model. As one insider close to Anthropic noted, the prevailing question isn’t “do I need the best intelligence for this task?” but rather, “Am I the kind of person who needs mid-level intelligence, or the best?” Many highly paid professionals across finance, politics, and technology are increasingly opting for the “best AI model,” a title Claude is actively vying for.
To maintain this reputation, Anthropic must continuously outpace its competition; early reactions to OpenAI’s GPT-5.6 Sol suggest this will be a significant challenge. While competitors like OpenAI and Google might explore revenue generation through advertising in their free and lower-cost tiers, Anthropic’s firm anti-advertising stance means higher prices might be its most viable short-term solution. In essence, the “golden era” of heavily subsidized consumer AI subscriptions appears to be drawing to a close.
Source: Wired – AI
