Claude Fable 5 is Anthropic's most capable generally available model as of June 2026 — built for demanding reasoning and long-horizon agentic work. This guide covers specs, API integration, pricing, and practical workflows.
Key specs
Context: 1M tokens by default. Output: up to 128K tokens per request. Thinking: adaptive thinking is always on when the thinking parameter is unset — use the effort parameter to control depth. Pricing: $10 per million input tokens, $50 per million output tokens.
What Fable 5 is best for
Multi-step coding agents, long document analysis, programmatic tool calling, task budgets, memory tools, and code execution. It powers the next generation of Claude Code workflows where the model plans, executes, and iterates across an entire repository.
Refusals & production integration
Unlike Claude Mythos 5, Fable 5 includes safety classifiers that can decline requests with stop_reason: refusal (HTTP 200, not an error). Production apps need fallback logic — server-side fallbacks, SDK middleware, or manual retry on another Claude tier. Anthropic offers fallback credit so you don't pay prompt-cache costs twice.
Fable 5 vs Opus 4.8
Opus 4.8 remains stable for teams mid-migration. Fable 5 adds 1M context (vs 200K+), task budgets, compaction, and stronger agentic tooling. Anthropic's migration guide recommends Fable 5 for new agent pipelines.
Getting started
Model ID: claude-fable-5. Available on Claude API, AWS Bedrock, Google Vertex AI, and Microsoft Foundry. See our flagship models comparison and Opus → Fable migration guide.
