Anthropic's New Claude Models Are Cheaper and Less Likely to Refuse Your Questions

Key points

• Anthropic launched Claude Fable 5.1 and Mythos 5.1, two updated AI models, in mid-2025.
• Fable 5.1 costs roughly 25 percent less than its predecessor, and up to 45 percent less for complex multi-step tasks.
• Fable 5.1 is available to all users now; Mythos 5.1 is limited to testers in a preview programme called Project Glasswing.
• A new enterprise privacy option will store customer data on the customer's own cloud servers, not Anthropic's, rolling out later this year.

Anthropologic has spent months hearing the same complaints from business customers: the models cost too much, they remember things they shouldn't, and they refuse to answer perfectly reasonable questions. The company's answer is two new AI models, meaning software tools you type questions or instructions into, called Claude Fable 5.1 and Mythos 5.1.

What's actually different this time?

Fable 5.1 delivers stronger results than Fable 5 while costing about 25 percent less on average. For complicated tasks where the model runs through many steps on its own, the saving jumps to around 45 percent. That price drop comes largely from cheaper handling of cached data, which is information the model already processed and stored so it doesn't have to repeat the work.

Early testers are impressed. Dan Shipper, chief executive of Every, said Fable 5.1 is "the strongest coding model we've used, but now it's fast, token-efficient, and crucially actually speaks like a normal person." Aaron Levie, chief executive of Box, reported that his company's automated assistant built on Fable 5.1 caught subtle details in data that the previous model missed on the same test.

Mythos 5.1 is the more powerful of the two, built for tasks that need deeper thinking. One early tester noted that Mythos 5.1 running on its lower-effort setting matches the performance of the older Mythos model running at full power, which matters because lighter settings cost less and respond faster.

Will it stop refusing perfectly ordinary questions?

Yes, partially. Fable 5.1 has what Anthropic calls "more precise safeguards," meaning it should be less likely to refuse basic biology questions that the previous model sometimes blocked. Mythos 5.1 keeps the same restrictions as before, so that change applies only to Fable.

Anthropologic is also loosening the rules around cybersecurity work. Fable 5.1 can now help identify weaknesses in software code, though tasks like attempting to break into systems or generating attack tools will still be redirected to a different, more restricted model called Opus.

What does this mean for businesses worried about privacy?

Businesses using Anthropic's enterprise tier will soon have an option called Enterprise Frontier Safeguards. It stores conversation data on the company's own cloud servers rather than on Anthropic's infrastructure, so the customer keeps full control. Anthropic says this offers "complete privacy." The feature starts rolling out later this autumn.

For ordinary users, Fable 5.1 is available right now on all of Anthropic's platforms. Mythos 5.1 is currently limited to participants in Project Glasswing, a closed testing programme, as first reported by The Verge AI.

Common questions

Do I need to do anything to get the cheaper pricing?

No action required. The lower prices for Fable 5.1 apply automatically if you already use Claude through Anthropic's website or API, which is the technical connection businesses use to plug Claude into their own software.

Can I join Project Glasswing to try Mythos 5.1?

Anthropologic has not publicly opened sign-ups for Project Glasswing. The programme is currently invite-only, and Anthropic has not announced a date for a wider release.