Claude Fable 5 pricing: Anthropic adds usage-based fees July 12

Anthropic will start charging usage-based token fees for Claude Fable 5 on July 12 at 11:59PM PT for subscribers to its $20, $100 and $200 monthly plans. The company set the rates at $10 per million tokens sent to Claude and $50 per million tokens the model generates in replies.

What exactly is changing for Claude subscribers?

Anthropic is moving the consumer version of Mythos 5, called Claude Fable 5, behind usage-based billing for paid plans: subscribers on the $20, $100 and $200 tiers will owe additional token charges after July 12 at 11:59PM PT. The company gave the concrete example that if a $20 subscriber sends one million tokens and receives one million tokens in responses, they would pay an extra $60, bringing their monthly total to $80.

The company previously offered Fable 5 to subscribers at no extra cost during an extended promotional period. Anthropic’s June 7 blog post said it expected demand for the model to be "very high, and difficult to predict," and the U.S. government briefly banned Fable 5 for foreign nationals before approving its general release on July 1.

How does this fit into broader AI pricing trends?

Usage-based billing has long been standard for developers using APIs, and Anthropic is applying those same rates to consumers, matching its API prices of $10 per million input tokens and $50 per million output tokens. The shift echoes recent moves elsewhere: AI coding startups such as Cursor pivoted from unlimited subscriptions to usage-based models, and Anthropic has already started charging large business customers based on employee usage rather than a flat fee.

Anthropic has cited capacity constraints as a factor. The company has struck "multibillion-dollar deals" for data center capacity with SpaceX, Amazon and Google, but says it still lacks the compute it wants. Anthropic spokesperson Reem Ateyeh said the company aims to return Fable 5 to subscription plans "when sufficient capacity allows," and intends to do so "as quickly as we can."

Nick Turley, OpenAI’s former head of ChatGPT now overseeing enterprise products, framed the broader issue plainly: "It’s possible that, in the current era, having an unlimited [AI] plan is like having an unlimited electricity plan," he said in a podcast earlier this year.

Why it matters

The change is a test of how much consumers will pay for higher-tier model access and how much usage they accept paying for directly. Claude’s consumer audience is growing: Sensor Tower data put Claude at 245 million unique visitors in May, more than double February levels, though it remains below ChatGPT’s 1.11 billion and Google Gemini’s 662 million monthly uniques. For heavy users and professionals who already run large API bills, token-based pricing will be familiar; for mainstream subscribers used to flat-fee access, it raises the cost and behavioral questions around when to use an advanced model.

Anthropic is also using product design to keep users engaged. A recent Claude feature called Reflect offers an analytics dashboard that visualizes what you ask Claude and how often, nudges users to manage their AI habits, and suggests product features like Projects to reduce repetitive context. Reflect is available in beta for Free, Pro and Max users with memory turned on and excludes health-integration conversations from insights.

What to watch

Look for how many subscribers hit measurable token usage that produces noticeable bills after July 12, and whether Anthropic restores Fable 5 to flat subscription tiers once capacity expands. Also watch competitive responses: OpenAI and Google are expected to pursue different revenue levers, including more advertising in lower tiers, while Anthropic is testing whether usage fees plus product hooks like Reflect keep users paying and loyal.