Anthropic has announced that Claude Fable 5.1 is now available in GitHub Copilot. ^[1] Anthropic launched the Claude Fable 5.1 AI model on the Harvey platform on September 1, 2026. ^[2] Claude Code v2.1.257 added a Containment Escape rule to auto mode, which prevents automatic approval of cloud metadata-credential fetches, egress evasion, and cross-tenant reach unless the environment marks them expected. ^[3] Claude Code v2.1.257 added Claude Fable 5.1 as the default Fable model, featuring 1M context and pricing of $10/$50 per million tokens with $0.25 per million token cache reads. ^[4] Claude Fable 5.1 is the latest model in Anthropic's Mythos class and is designed for long-horizon, autonomous coding and knowledge-work tasks. ^[5] Zero data retention access is available for certain eligible enterprise customers for Claude Fable 5.1. ^[6] Anthropic is rolling out its Enterprise Frontier Safeguards, allowing eligible customers to store data within their own infrastructure or on Harvey's systems under a time-limited exemption. ^[7] Early testing indicates that Claude Fable 5.1 excels in delivering quality results even at lower reasoning levels, using roughly half the tokens compared to higher settings. ^[8] Harvey clarified that the Claude Fable 5.1 model processes all data in the United States, with no regional processing options available. ^[9] Claude Code v2.1.257 fixed a security issue where plugins could read files outside their own directory through symlinked command, agent, skill, hooks, or component paths; such paths are now refused with an error. ^[10] Claude Code v2.1.257 added the CLAUDE_CODE_SUBAGENT_MODEL_FORCE environment variable to apply the subagent model to every subagent, ignoring per-spawn and agent-definition model overrides. ^[11]