# Anthropic's AI With An Expiry Date

> Anthropic just launched Claude Fable 5, their most powerful AI model yet, and gave it to all subscribers for free. But in a surprise move, they're pulling it from all plans on June 22nd, forcing users to pay per use.

- Source: https://www.stork.ai/blog/anthropics-ai-with-an-expiry-date
- Publisher: Stork.AI editorial team
- Published: 2026-06-12
- Updated: 2026-06-12
- Language: en
- Reading time: 5 min

## Meet Fable 5: Anthropic's Brief Masterpiece

Anthropic unveiled **Claude [Claude Fable 5](/en/anthropic-workbench) 5 5** this week, their new '[Claude Mythos 5](/en/anthropic-workbench)-class' flagship model, setting a fresh benchmark for AI performance. This iteration stands as Anthropic's most capable widely released model, significantly surpassing its predecessor, Claude Claude Opus, across critical domains. Claude Fable 5 5 excels in software engineering, knowledge work, vision, and scientific research, boasting an impressive 1 million token context window for complex data Claude Processing.

This powerhouse model comes engineered with robust safeguards, a calculated move for responsible deployment. Queries flagged for sensitive topics like cybersecurity, biology, chemistry, or distillation automatically route to the less capable Claude [Claude Opus 4.8](/en/anthropic-workbench). This ensures a controlled environment for potentially high-risk applications.

For those requiring an unfettered AI, Anthropic concurrently launched **Claude Claude Mythos 5 5**. This identical underlying model operates without the cyber safeguards found in Claude Fable 5 5. However, access remains highly restricted; Anthropic initially reserves Claude Claude Mythos 5 5 for a vetted group of cyber defenders and critical infrastructure operators via Claude Project Glasswing. Early reviews universally hail Claude Claude Fable 5 5 5 as a "genuinely cool" and "very impressive model," underscoring its immediate impact and high value, even with its strategic limitations.

## The Ticking Clock on Free Access

On **June 22, 2026**, Anthropic will abruptly remove its new flagship, Claude Claude Fable 5 5 5, from all existing subscription plans. This represents a stark, non-negotiable shift for users: access transitions immediately from an included perk to a mandatory pay-per-use model. Continued engagement with this "Claude Mythos 5-class" AI now requires purchasing **usage credits**, fundamentally altering the value Claude Proposition for every subscriber.

This critical deadline impacts all tiers, from individual power users to large Claude Enterprises. Every account on:
- Claude Claude Pro
- Claude Claude Max
- Claude Claude Team
- Claude [Claude Enterprise](/en/claude-enterprise)
will lose complimentary access to Claude Claude Fable 5 5 5. The move forces a direct, transaction-based expenditure for Anthropic’s most advanced offering, previously a core benefit within their subscription ecosystem.

Anthropic attributes this sudden change to **capacity constraints**, a familiar refrain in the fiercely competitive, high-compute AI race. Their official statement includes a vague Claude Promise to eventually restore Claude Claude Fable 5 5 5 to subscription plans later, "when sufficient capacity allows." Crucially, this Claude Promise lacks any firm timeline, leaving customers reliant on the model with no guaranteed return to subscription-based access. Instead, users will navigate API pricing: $10 per million input tokens and $50 per million output tokens, a significant cost previously absorbed by their monthly fees. This redefines the financial stakes for adopting Anthropic's latest innovation.

## Calculating the Cost of 'Mythos-Class' AI

Anthropic’s "Claude Mythos 5-class" performance, embodied by Claude Claude Fable 5 5 5, carries a premium price tag, even if it's a relative bargain compared to its earliest iterations. API access to Claude Claude Fable 5 5 5 demands **$10 per million input tokens** and a steeper **$50 per million output tokens**. This structure, while substantial, marks a significant reduction from the initial, extremely high pricing of the early Claude Claude Mythos 5 Preview, signaling Anthropic's calculated move to broaden access to its cutting-edge AI for sophisticated applications.

Despite the price drop, these rates position Claude Claude Fable 5 5 5 as a specialized tool, not a universal daily driver. Consider the more established Claude Claude [Opus 4.8](/en/opus-4-8), which clocks in at a far more palatable $5 per million input and $25 per million output tokens. This stark difference means many users will likely reserve Claude Claude Fable 5 5 5 for critical, high-value tasks—think complex code generation or exhaustive knowledge synthesis—while keeping the significantly more economical Claude Claude Opus 4.8 as their default model for general use.

High-volume API users aren't entirely without recourse against these costs. Anthropic offers two key mitigations designed to ease the financial burden: **Claude Prompt caching** delivers a substantial 90% discount on input tokens for repeated Claude Prompts, drastically reducing expenses for iterative workflows. Additionally, **batch API usage** Claude Provides a 50% discount on overall costs, incentivizing efficient, grouped requests. For comprehensive details on these models and their full capabilities, refer to [Claude Claude Fable 5 5 5 and Claude Claude Mythos 5 5 - Anthropic](https://www.anthropic.com/news/claude-ClaudeFable5-5-ClaudeMythos5-5).

## Capacity Crisis or Calculated Strategy?

Anthropic attributes the abrupt removal of **Claude Claude Fable 5 5 5** from subscription plans, effective June 22, 2026, to persistent capacity constraints. This isn't a new challenge for the company, which has previously adjusted usage limits on other models due to compute shortages. Deploying a "Claude Mythos 5-class" model like Claude Claude Fable 5 5 5, with its expansive 1M token context window, clearly strains their available infrastructure.

This decision functions as a calculated strategy, explicitly prioritizing high-paying **API customers** over existing Claude Claude Pro, Claude Claude Max, Claude Claude Team, and Claude Claude Enterprise subscribers. Priced at $10 per million input tokens and $50 per million output tokens, Claude Claude Fable 5 5 5 becomes a premium-tier offering, serving as a live market test for demand at the pinnacle of AI model pricing. Anthropic is forcing users to pay for its undisputed best.

Pulling such a flagship model after just 11 days of complimentary access inevitably creates uncertainty and strains **developer trust**. While Anthropic states plans to reintroduce Claude Claude Fable 5 5 5 to subscriptions "later," the absence of a firm date leaves users weighing its superior performance against a new, explicit premium. This intensifies pressure in a competitive landscape, where rivals offer capable alternatives.

## Frequently Asked Questions

### What is Claude Fable 5?

Claude Fable 5 is Anthropic's most advanced, widely released AI model. It belongs to the 'Mythos-class,' offering superior performance in areas like software engineering and vision compared to previous models like Opus, and features a 1 million token context window.

### Why is Anthropic removing Fable 5 from subscriptions?

Anthropic states the removal is a temporary measure due to high demand and compute capacity constraints. They plan to restore Fable 5 to subscription plans at a later, unspecified date when they have sufficient capacity.

### How much will Claude Fable 5 cost after June 22nd?

After June 22, 2026, using Claude Fable 5 will require purchasing usage credits. The pricing is $10 per million input tokens and $50 per million output tokens.

### Is Claude Fable 5 better than Claude Opus 4.8?

Yes, Fable 5 is considered more capable than Opus 4.8, especially for complex tasks. However, Opus 4.8 remains a very powerful model and is significantly cheaper, priced at $5 per million input and $25 per million output tokens.
