Claude Fable 5.1 Frontier Model
Anthropic released Claude Fable 5.1 and Claude Mythos 5.1 on 1 September 2026: the same model in two safeguard tiers, with Fable 5.1 generally available and Mythos 5.1 restricted to vetted cybersecurity and life-sciences organisations. Built for long-horizon agentic coding, multistep research and document, spreadsheet and slide work, with cache reads 75% cheaper
Tool Interface
Interactive tool will be available soon
Features
- ✓ One underlying model, two safeguard tiers: Fable 5.1 is open to all, Mythos 5.1 goes through vetting
- ✓ Long-horizon agentic coding: features spanning a whole codebase, code review, performance work and multi-day autonomous sessions
- ✓ Multistep research plus document, spreadsheet and slide output for knowledge work
- ✓ Input and output prices match Fable 5, while cache reads drop to a quarter, at $0.25 per million tokens
- ✓ 1M-token context window and up to 128K output tokens, with adaptive thinking always on at a default high effort
How to Use
- Read the official launch post for capabilities and the safeguard split: https://www.anthropic.com/claude-fable-and-mythos-5-1
- Switch calls to claude-fable-5-1 per the official model docs and trial it in Claude Code or Claude on real work
- Work through the three documented breaking changes for compatibility, then run shadow traffic and a narrow canary
- Design for cache reuse by default so the 75% cache-read cut actually lands on your bill
FAQ
How do Fable 5.1 and Mythos 5.1 differ?
In Anthropic's own words they are the same model with different levels of safeguards. Fable 5.1 is generally available, while Mythos 5.1 keeps lighter guardrails but is limited to vetted cybersecurity and life-sciences organisations. Official post: https://www.anthropic.com/claude-fable-and-mythos-5-1
When was it released?
Anthropic released it on 1 September 2026. Per the Claude Platform model docs, Fable 5.1 is Active (latest) with retirement no sooner than 1 September 2027.
What changed on pricing?
Input and output pricing matches Fable 5, but cache reads fall to $0.25 per million tokens, a quarter of the previous $1 rate. For context-heavy agents and long sessions that line item is usually where the money actually goes.
What is it best at?
Anthropic positions it for ambitious coding projects: work spanning an entire codebase, code review, performance work and multi-day autonomous sessions, alongside multistep research and document, spreadsheet and slide output. The docs suggest starting most workloads on Claude Opus 5 and moving to Fable 5.1 when Opus 5 at higher effort still falls short.
What should I watch when upgrading?
Three breaking changes affect existing Fable 5 integrations: forced tool use returns an error, earlier models cannot read its thinking blocks, and editing earlier turns invalidates thinking blocks. Regress against each before rolling out.
Where can I use it?
Per the platform docs it is available through the Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry and Claude Platform on AWS, and it landed in Claude Code on launch day.