Claude Fable 5.2 (Max, Default Fallback)

Claude Fable 5.2 is Anthropic's generally available, safeguarded deployment of the same underlying model as Claude Mythos 5.2 (verification-program access only). It targets demanding reasoning and long-horizon agentic work with text and image input, text output, tool use, multilingual and vision support, and adaptive thinking that is always on (Claude API default effort high). Context window is 1M tokens with 128K max output on the sync Messages API; reliable knowledge and training-data cutoffs are June 2026. First-party pricing drops to $8/$40 per million input/output tokens (from $10/$50 on Fable 5.1), with cache reads at $0.20 (0.025x base input), 5-minute cache writes at $10 and 1-hour cache writes at $16 per million tokens; Batch is 50% off. Optional fast mode gives up to 2.5x higher output tokens per second at $16/$80 per million input/output tokens and is not available with the Batch API. Anthropic reports outputs up to 30% faster than Fable 5.1 and, in early testing across claude.ai, Claude Code and Cowork, about 55% fewer cyber and bio safeguard blocks than Fable 5.1. Self-reported launch scores with safeguards enabled include FrontierCode 2.0 49.1% (xhigh), FrontierSWE v2 69.9%, SWE-bench Pro 93.0%, Terminal-Bench 4.0 67.6%, Humanity's Last Exam 74.2% with tools, OSWorld 2.1 offline set 89.3% partial, and AutomationBench 51.4%. Available as `claude-fable-5-2` on the Claude API, Amazon Bedrock (`anthropic.claude-fable-5-2`), Claude Platform on AWS, Claude Models on Google Cloud and Microsoft Foundry.