Model Comparison
Fable 5 vs Claude Opus: What's Actually Different
Two Anthropic model lines, built for different depths of work.
Fable and Claude Opus are two of Anthropic's model lines, and they are built for different jobs. Fable 5 launched on June 9, 2026 as a Mythos-class model made safe for general use, and Claude Fable 5.1 succeeded it on September 1, 2026. On the Opus side, Opus 4.8 was followed by Opus 5 and then Claude Opus 5.5, released September 22, 2026. This page covers what separates the two lines, where they cooperate, and how to choose.
Fable is Anthropic's top general-use line: Claude Fable 5.1 (September 1, 2026) is its most capable model open to all customers, built for demanding reasoning and long-horizon agentic work. Opus is faster, and Anthropic's docs suggest Claude Opus 5.5 (September 22, 2026) as the starting point for most workloads. The lines also cooperate: at launch, Fable 5 routed classifier-flagged cybersecurity, biology and chemistry, and distillation requests to Opus 4.8, and more than 95% of Fable 5 sessions involved no fallback.
Published · Last reviewed
Free lesson · no account neededMeet ClaudeWhy Claude?6 minStart the free lessonSide by Side
How Fable and Claude Opus differ
Both models come from Anthropic, but they are optimized for different kinds of work.
Fable: Mythos class, made general
Anthropic calls models above its Opus class Mythos-class. Fable 5 (June 9, 2026) was the first one made safe for general use, and Fable 5.1 (September 1, 2026) is the current Fable model, aimed at demanding reasoning and long-horizon agentic work.
Claude Opus: the starting point
Claude Opus 5.5 (September 22, 2026) is the current Opus model, built for long-running agentic coding and knowledge work. Anthropic's docs suggest starting most workloads with it. Opus 5 and Opus 4.8 remain available as legacy models.
Speed versus depth
Anthropic rates Fable 5.1 as slower and Opus 5.5 as moderate in comparative latency. The docs suggest moving from Opus 5.5 to Fable 5.1 when evals at higher effort still fall short on demanding, long-horizon tasks.
Fable 5 API pricing
Fable 5 is priced at $10 per million input tokens and $50 per million output tokens. Consult Anthropic's pricing page for current Fable 5.1 and Opus rates.
Built-in safety routing
At launch, Fable 5 used classifiers that detect sensitive queries and route them to Claude Opus 4.8. More than 95% of sessions involved no such handoff. For Fable 5.1, Anthropic says some life sciences and dual-use cybersecurity requests still go to its Opus models.
One family, two lines
Fable and Opus are both Claude models. Fable sits above Opus in capability, and Opus sits above Sonnet and Haiku. For every Claude release with its date, see the Claude Model Timeline page on Nightschool AI.
How They Cooperate
Safety routing: when Fable 5 hands off to Opus
One of the less obvious aspects of Fable is that it does not always operate alone. When Fable 5 launched, classifiers watched for signals in three sensitive areas: cybersecurity, biology and chemistry, and distillation attempts. When a classifier fired, the request received a response from Claude Opus 4.8 instead.
Anthropic reported that more than 95% of Fable 5 sessions involved no fallback, meaning the handoff was designed for edge cases rather than everyday use. With Fable 5.1, Anthropic says queries related to life sciences research and development, and several kinds of dual-use cybersecurity tasks, are still directed to its Opus models.
This cooperation matters for developers building on the API. If your application touches any of the flagged areas, you may receive Opus responses from a Fable request. Knowing the routing exists helps you interpret latency differences and response characteristics when they appear.
Which to Choose
Picking the right model for your task
Choose Fable 5.1 when the work is the hardest and longest you have: agent sessions that run for hours, multistep deep research, or analysis carried through to a finished document, spreadsheet, or deck. Anthropic positions it as the highest available capability for customers without special access.
Choose Claude Opus 5.5 for most other demanding work: multihour coding agents, large refactors, complex systems engineering, and vision-heavy workflows. It is faster than Fable, it is where Anthropic's docs suggest most workloads should start, and it handles sensitive areas without a classifier handoff.
For teams building products, the two models are not mutually exclusive. A common pattern is to run most work on Opus 5.5 and send only the requests that fall short to Fable 5.1. Understanding both models' strengths lets you design workflows that use each where it performs best.
Learn to work with Claude's flagship models, hands-on
Reading a comparison and actually knowing how to prompt, chain, and deploy these models are different skills. The free Nightschool AI curriculum is practical from the first lesson, no theory-only modules.
Frequently Asked Questions
When did Fable 5 come out?
Fable 5 was released on June 9, 2026. Its successor, Claude Fable 5.1, was released on September 1, 2026, and Fable 5 is now a legacy model that is still available.
What does Mythos class mean?
Anthropic uses Mythos-class for models that sit above its Opus class in capability. The first was Claude Mythos Preview, released in April 2026 only to Project Glasswing partners. Fable 5 was the first Mythos-class model made safe for general use. Models named Mythos, such as Mythos 5.1, are the same model as the matching Fable with some safeguards lifted, offered only to verified organizations.
What is Fable 5's API pricing?
Fable 5 is priced at $10 per million input tokens and $50 per million output tokens through Anthropic's API. These figures come from Anthropic's official release announcement at anthropic.com/news/claude-fable-5-mythos-5.
Why does Fable 5 route some queries to Claude Opus 4.8?
Fable 5 includes a built-in classifier that detects queries in sensitive domains: cybersecurity, biological and chemical information, and knowledge distillation. When the classifier flags a query, Fable 5 routes it to Opus 4.8 rather than answering directly. This is a safety design. More than 95% of sessions involve no such routing.
Is Fable better than Claude Opus?
Fable is the more capable line. Anthropic calls Fable 5.1 its most capable model open to all customers. Opus 5.5 is faster, and Anthropic's docs suggest it as the starting point for most workloads, moving to Fable 5.1 when Opus at higher effort still falls short. The right pick depends on how demanding and how long the task is.
Can I use Fable 5 and Claude Opus 4.8 together?
Yes. Fable already relies on Opus models for some sensitive requests through its safeguards. You can also design your own pipelines that run most work on Opus 5.5 and escalate only the hardest, longest tasks to Fable 5.1.
What sensitive domains trigger the Fable 5 handoff?
At launch, Fable 5's classifiers covered three categories: some cybersecurity topics, most biology and chemistry requests, and requests flagged as distillation attempts. Those were routed to Claude Opus 4.8. For Fable 5.1, Anthropic says life sciences research and development queries and several kinds of dual-use cybersecurity tasks still go to its Opus models.
Are Fable 5 and Opus 4.8 still available on the API?
Yes, as legacy models. Anthropic's current models are Claude Fable 5.1 (claude-fable-5-1) and Claude Opus 5.5 (claude-opus-5-5), alongside Sonnet 5.5 and Haiku 5.5. Fable 5 and Opus 4.8 remain available, each with a published retirement window, so check Anthropic's model deprecations page before building on one.
Put both models to work
The best way to understand the difference is to use them.