Claude Fable 5.2 and Claude Mythos 5.2 improve coding, reasoning, and vision work, while Fable generates outputs up to 30% faster than Fable 5.1. Anthropic positions both models for demanding, open-ended technical tasks; they share the same underlying model, but Fable includes cyber and biology safeguards.
Stronger technical work with Fable and Mythos
Teams can use Fable 5.2 for harder coding, reasoning, and visual-analysis tasks. It can act as an Advisor, where other Claude models consult it on difficult work, or as an Orchestrator that delegates simpler tasks to smaller models. At Medium effort, the default in Claude apps and Claude Code, Anthropic says Fable 5.2 can complete challenging tasks with fewer tokens than Fable 5.1.
Cybersecurity work and Claude Security
Vetted organizations can use Mythos 5.2 for vulnerability discovery, reproduction, and remediation. Anthropic reports stronger vulnerability identification and judgment than earlier Mythos models. Access to Mythos cyber capabilities requires the Cyber Verification Program.
Claude Enterprise customers can use Claude Security, now including Mythos 5.2. Its recent update can monitor repositories as code changes, track findings until they are fixed, incorporate further context, and connect to user tools through API and MCP.
Fable safeguards reduce unnecessary blocks
Fable 5.2 keeps classifiers for high-risk cybersecurity activity and most research-biology tasks while reducing unnecessary blocks. Across claude.ai, Claude Code, and Cowork, Anthropic observed about 55% fewer blocks than with Fable 5.1.
Fable blocks penetration testing, exploit generation, binary-based vulnerability work, and most research-biology requests. Blocked cyber requests receive an Opus 4.8 response; blocked biology requests receive an Opus 5 response. For a small set of AI research topics, Fable and Mythos also use transparent Opus 5 fallbacks.
Preserved thinking for API users
Preserved thinking now applies to all accounts. The API feature prevents users from editing prior Claude context to extract reasoning. Developers should consult Anthropic's preserved thinking documentation when updating integrations for the new API features.
Enterprise Frontier Safeguards private beta
Enterprise Frontier Safeguards is in private beta. It runs automated safety monitoring on prompts and responses stored in customer infrastructure, making it compatible with zero data retention. Eligible Enterprise customers can continue using Fable 5.2 and Mythos 5.2 with zero data retention before implementing safety monitoring. Other customers have 30-day retention for that monitoring.
Availability and optional fast mode
Fable 5.2 is available today on all platforms, including Amazon Web Services, Google Cloud, and Microsoft Azure. Developers can use claude-fable-5-2 on the Claude Platform. Security professionals and life-sciences researchers can apply for Mythos 5.2 through the Cyber Verification Program and Life Sciences Verification Program.
Optional fast mode is available in Claude Code, the Claude Platform, Claude Models on Google Cloud in Anthropic-managed and Google-managed configurations, and Anthropic-hosted Microsoft Foundry. It can increase token generation speed by up to 2.5 times. Mythos fast mode is only available to organizations in the Verification Programs.
What this means for your team
Developers can use Fable 5.2 on difficult code changes and technical investigations. QA engineers can apply the same model to test and review work, while verified AppSec teams can investigate and remediate findings with Mythos 5.2. Claude Security gives Enterprise teams a way to follow findings as repositories change, using shared code context across engineering and security.
Based on the official Anthropic release notes.
How Merito helps
Merito helps teams adopt and govern AI-assisted engineering workflows through AI-Powered Software Development and Managed AI Operations. We deliver Anthropic Claude strategy, rollout and operations services, and US public-sector buyers can order Claude licenses through Merito on GSA MAS contract 47QSWA18D008F. Our Anthropic Claude team helps you put the Fable and Mythos 5.2 changes to work across QA, AppSec, and development, with AI rollout and activation so every role gets the same agentic tooling and AI readiness and governance to keep that access controlled.
Learn more about Anthropic, AI rollout activation, and AI readiness and governance at www.merito.com.