Anthropic — June 2026

Claude Fable 5: The Next Generation of Intelligence

A Mythos-class model built for the hardest knowledge work and coding problems. State-of-the-art on nearly all tested benchmarks of AI capability — exceptional in software engineering, knowledge work, vision, and scientific research.

Creator
Anthropic
Launch
June 2026
Class
Mythos-class
Pricing
$10 / $50 per MTok
Capabilities

What makes Claude Fable 5 different

Claude Fable 5 introduces Anthropic's fifth model generation for your most ambitious work. It can work autonomously for longer than any previous Claude model.

Software Engineering

Fable 5 compresses months of engineering into days. Stripe reported a codebase-wide migration in a day that would have taken a team over two months by hand. Highest scorer on Cognition's FrontierCode evaluation.

🔍

Knowledge Work

Strongest performance on complex analytical tasks. Highest score on Hebbia's Finance Benchmark for senior-level reasoning with substantial gains in document reasoning, chart interpretation, and problem solving.

👁

Vision

New state-of-the-art for vision tasks. Extract precise numbers from scientific figures, rebuild web apps from screenshots, and play games like Pokémon FireRed with a minimal vision-only harness.

🧠

Memory & Long Context

Stays focused across millions of tokens. Persistent file-based memory improved performance three times more than for Opus 4.8 in long-running tasks like the deck-building game Slay the Spire.

How It Works

Autonomous, thorough, and self-validating

Fable 5 is thorough, proactive, and tests its own work. Run it in an agent harness like Claude Code and it can work for days at a time: planning across stages, delegating to sub-agents, and checking its own outputs.

01

Plan across stages

Fable 5 breaks complex projects into phases, allocates resources, and prioritizes the work that matters most.

02

Delegate to sub-agents

It spins off sub-tasks, monitors progress, and re-allocates effort when it spots a better path forward.

03

Write tests & validate

It writes its own tests to check its work, uses vision to compare outputs against designs, and self-corrects.

04

Deliver completed work

Teams hand off large projects and review finished deliverables rather than supervising every step.

Use Cases

Built for ambitious, long-running work

Try problems you weren't able to solve with other models. Fable 5 is thorough, proactive, and tests its own work.

Agents

Run Claude Fable 5 in an agent harness like Claude Code or Claude Managed Agents, and it can work for days at a time: planning across stages, delegating to sub-agents, and checking its own work.

Coding

Our most capable model for ambitious coding projects, including large migrations, complex implementations, and multi-day autonomous sessions. It can write its own tests to check its work, implement designs with high fidelity, and use vision to check outputs against goals.

Enterprise Workflows

Handles complex, multi-stage knowledge work with minimal oversight, from deep research and analysis to deliverables ready for your review. Teams can hand off large projects and review completed work rather than supervising every step.

Scientific Research

Fable 5 excels at life sciences research, positing novel hypotheses and accelerating drug design. Using Mythos 5, protein design experts accelerated aspects of the drug design process by around ten times.

Pricing & Availability

Accessible pricing for frontier intelligence

Claude Fable 5 is available on the Claude API, consumption-based Enterprise plans, and through Amazon Web Services, Google Cloud, and Microsoft Foundry.

$50
per million output tokens
US-only inference at 1.1x pricing available

Available Everywhere Today

Developers can use claude-fable-5 via the Claude API. Fable 5 is fully available on consumption-based Enterprise plans and the Claude API from today.

Get API Access
Safeguards

Launched with robust protections

Fable 5 includes safety classifiers that detect potential misuse across cybersecurity, biology and chemistry, and distillation. Flagged queries are automatically routed to Opus 4.8 — a far better experience than an outright refusal.

Cybersecurity

Classifiers prevent Fable from providing uplift on offensive cyber tasks. More than 95% of sessions involve no fallback at all.

Biology & Chemistry

Dual-use capabilities are protected by broad safeguards. Bioweapons-related queries and advanced biological research queries fall back to Opus 4.8.

Distillation Protection

Classifiers detect large-scale extraction attempts to prevent proliferation of near-frontier AI capabilities without appropriate safeguards.

Feedback

What early customers are saying

Customers with early access ran their own tests on Fable 5. Here's a selection of what they're seeing.

Claude Fable 5 is the state of the art model on CursorBench. It's opened up a class of long-horizon problems that were out of reach for earlier models.
Michael Truell — CEO and Co-founder, Cursor
Claude Fable 5 is a real step forward for the developers GitHub serves. It took on complex, long-horizon coding tasks with a level of autonomy and reliability that exceeded previous benchmarks.
Mario Rodriguez — Chief Product Officer, GitHub
Claude Fable 5's reasoning is a clear step beyond Opus 4.8. It works at senior research scientist grade — picking directions, allocating resources, killing its incorrect beliefs, and producing novel first-principles outputs.
Sean Ward — CEO and Co-founder, AlphaSignal
Claude Fable 5 is the highest-scoring model on FrontierBench, Cognition's frontier coding eval. It excels at long-horizon reasoning and generalizes to unfamiliar tools out of the box.
Scott Wu — CEO, Cognition
FAQ

Frequently asked questions

The fastest answers to the questions people ask first about Claude Fable 5.

When should I use Claude Fable 5?

We offer Claude models across the spectrum of speed, price, and performance. Claude Fable 5 is our most capable generally available model. It is best for ambitious, long-running, asynchronous work that previous models could not sustain.

How does Fable 5 compare to Opus 4.8?

Fable 5 is a Mythos-class model that substantially exceeds Opus 4.8 across nearly all benchmarks — coding, knowledge work, vision, scientific research, and long-context tasks. The longer and more complex the task, the larger Fable 5's lead.

What safeguards does Fable 5 have?

Fable 5 comes with classifiers that detect potential misuse in cybersecurity, biology and chemistry, and distillation. Flagged queries automatically receive a response from Opus 4.8 instead. More than 95% of sessions involve no fallback.

How much does it cost to use Claude Fable 5?

Fable 5 is priced at $10 per million input tokens and $50 per million output tokens, with the existing 90% input token discount for prompt caching. US-only inference is available at 1.1x pricing.

How do I access Claude Fable 5?

Developers can use claude-fable-5 via the Claude API. Fable 5 is available on consumption-based Enterprise plans, and through Amazon Web Services, Google Cloud, and Microsoft Foundry.

Primary Sources

Verified information and documentation

Every claim on this page is grounded in official Anthropic documentation and announcements so you can verify the details yourself.