The Plain-English Explanation
Anthropic’s Claude family has four tiers: Haiku (fast and cheap), Sonnet (the everyday default), Opus (the flagship for hard work) and, since June 2026, a Mythos class that sits above all of them. Fable is the public face of that class. Fable 5 launched on 9 June 2026, was suspended for three weeks under a US export-control directive, and was restored on 1 July. Fable 5.1 replaced it on 1 September 2026.
The Fable/Mythos split is the interesting part. They are the same model. Fable 5.1 ships with safeguards tuned for general use; Mythos 5.1 relaxes some of those safeguards for organisations doing legitimate cybersecurity and life-sciences work, under a trusted-access programme. It is the first time a frontier lab has openly sold one model at two safety settings.
Why It Matters
For most businesses, Fable 5.1 is the model you reach for when Opus 5 is not enough: very long, multi-step agentic tasks, deep research, and document, spreadsheet and slide work where getting it right matters more than getting it fast. It costs US$10 per million input tokens and US$50 per million output — double Opus 5 — but Anthropic cut cached-input pricing by 75%, so typical workloads land about 25% cheaper than Fable 5, and heavily agentic ones up to 45% cheaper.
It is also a lesson in vendor risk. The June suspension took the world’s most capable model offline for every customer overnight. Anyone building on a frontier tier should have a fallback model wired in and tested.
How It Works
On Claude.ai, Fable 5.1 is included on Max, Team Premium and premium Enterprise seats, and available on Pro and Team Standard through usage credits. In the API it is claude-fable-5-1, with a 1M-token context window and a per-request effort setting so you can trade depth for cost. Anthropic recommends Sonnet 5 for everyday work, Opus 5 for hard agentic and coding tasks, and Fable 5.1 for the small share of work where the extra capability pays for itself.
Examples in Practice
- A consulting team letting Fable 5.1 run a two-hour research task across 40 source documents and return a fully cited briefing pack.
- A software team using Fable 5.1 in Claude Code to trace a production bug to its root cause across a large codebase rather than patching the symptom.
- A finance team using Fable 5.1 to rebuild a 30-tab spreadsheet model with documented assumptions, then Sonnet 5 for the day-to-day edits.
Common Misconceptions
Myth: Mythos is a different, more powerful model than Fable.
Reality: They are the same weights. Mythos 5.1 simply runs with fewer safeguards for approved cyber and life-sciences partners.
Myth: You should use Fable for everything.
Reality: Most tasks do not need it. Sonnet 5 and Opus 5 are faster and cheaper, and on routine work you will not see a difference.
Myth: Frontier access is guaranteed once you pay for it.
Reality: The Fable 5 suspension in June 2026 showed that regulatory action can remove access overnight. Build with fallbacks.
Related Terms
Learn Mythos-Class Model (Claude Fable) in Depth
Claude AI: The 3-Hour Intensive walks business professionals through the whole Claude family — how to pick the right tier, Cowork, Skills, connectors and governance — with the Fable story as a live case study.
Explore Claude AI: The 3-Hour Intensive