Anthropic just launched its most powerful Claude ahead of its IPO — here's why it's telling users to stick with Opus
Latest from Tom's Guide ·
null
Anthropic has a new AI model, and it's more powerful than the Claude most people are using for chat. And that's precisely why the company doesn't actually recommend switching to it for everyday tasks.
Claude Fable 5.2 , released Monday alongside the more restricted Mythos 5.2, is designed for some of the toughest work AI can handle, including advanced coding, scientific research and complex engineering projects.
The timing is significant. Anthropic is also preparing for a massive initial public offering , putting fresh attention on how the company plans to turn increasingly powerful AI into a profitable business.
For Claude users, though, the immediate question is simpler: what can Fable 5.2 actually do, and is it worth using instead of Opus?
Fable 5.2 is designed for projects that require more than generating an email, summarizing a document or answering a question.
Anthropic says its newest model can handle complicated technical work that would otherwise require extensive human effort. During prerelease testing, its models helped engineers complete difficult software migrations, design working electronic devices and tackle advanced physics problems.
One of the more interesting applications is using Fable as an AI advisor. Rather than asking Fable to handle every task, developers can have smaller Claude models consult it when they encounter a particularly difficult problem.
Fable can also act as an orchestrator, coordinating smaller models and assigning them simpler tasks while it handles the more complicated work.
For everyday users, the most practical applications could include analyzing complex research, working through difficult technical problems or managing projects that involve multiple steps and large amounts of information.
Here's where Anthropic's lineup gets confusing. Despite introducing Fable 5.2 as one of its most capable models, Anthropic recommends that most people continue using Claude Opus 5.5 for everyday coding and knowledge work.
The company describes Opus as its strongest daily driver, while Fable is intended for more ambitious technical challenges. That means someone using Claude for writing, research, brainstorming or routine coding probably doesn't need to switch.
Fable becomes more interesting when a task requires sustained reasoning, complicated problem-solving or coordination between multiple AI systems.
There are also cheaper models in the lineup. Claude Sonnet 5.5 and Haiku 5.5 remain options for simpler work where speed and efficiency matter more than maximum capability.
The takeaway is that the most powerful Claude isn't automatically the best Claude for every job.
Anthropic has reduced the price of Fable compared with its previous version, but it remains more expensive than Opus.
Through the API, Fable 5.2 costs $8 per million input tokens and $40 per million output tokens. That's 20% less than Fable 5.1, although it's still twice the per-token price of Opus 5.5.
Anthropic also says the new model generates responses up to 30% faster than its predecessor. An optional fast mode offers even higher output speeds, but at twice the standard API token price.
For developers and businesses running large numbers of AI requests, those differences can add up quickly. If you run through your usage limits quickly, you might want to choose Opus 5.5 first.
For people using Claude through a subscription , access and usage limits also matter. Fable is available through supported Claude plans, although how much you can use depends on your subscription and any applicable usage credits.
Anthropic's latest model launch arrives at a particularly important moment for the company.
According to Reuters , Anthropic circulated IPO pricing information to banks before retracting it, saying the figures were inaccurate. The withdrawn proposal outlined a potential offering worth as much as $69.3 billion, although those numbers should not be treated as finalized.
The company is entering the public-market spotlight at a time when investors are scrutinizing the enormous cost of developing and operating advanced AI.
So, instead of offering one extremly powerful version of Claude for everyone, Anthropic is building a lineup designed for different types of work and different budgets.
Haiku and Sonnet can handle simpler, high-volume tasks. Opus serves as the everyday workhorse. Fable targets more demanding projects, while Mythos offers specialized capabilities to approved organizations working in fields such as cybersecurity and biomedical research.
It's a strategy that could help Anthropic serve more customers without requiring everyone to pay for its most expensive technology.
But it also creates a challenge: users need to understand why they should pay more for one model when another may already do everything they need.
As Anthropic moves toward its IPO, demonstrating that customers will pay for those additional capabilities could become just as important as achieving another impressive benchmark score.
If you're already using Claude Opus 5.5 for writing, research, brainstorming or everyday productivity, there's little reason to rush into Fable 5.2.
The new model is worth exploring if you're working on particularly difficult technical projects or tasks where existing AI models struggle to deliver reliable results.
For everyone else, Opus remains Anthropic's recommended starting point.