C

Models

Claude Fable 5.1

Anthropic's most capable generally available model, released September 1, 2026, for long-running agentic coding, multistep research, and document work. 1M-token context, 128K output, adaptive thinking that is always on.

Highlights

  • 1M context, 128K output

    Whole repositories and long agent transcripts fit in one window; effort can change mid-conversation without invalidating the cache.

  • Adaptive thinking, always on

    There is no thinking-off mode. The effort parameter controls depth, from routine edits to days-long tasks.

  • Built for long-horizon agents

    Stronger on tasks that run for hours across many files and tools, plus spreadsheets, documents, and slides.

Availability

Model id claude-fable-5-1 on the Claude API, Amazon Bedrock, Google Cloud, and Microsoft Foundry. In Claude for Pro, Max, Team, and Enterprise. Requires 30-day data retention.

Pricing

Paid
Input $10 / MTok
Output $50 / MTok
Cache read $0.25 / MTok
Same input and output prices as Fable 5; cache reads cost 75% less, which cuts typical workloads by about 25% and agentic ones by up to 45%. Batch API is half price. Claude Mythos 5.1 is the same model with looser safeguards for trusted-access programs.
#model
Check out site (opens in new tab)