Anthropic Covered Models: Fable, Mythos and Extra Safeguards
⚡ TL;DR Covered Models are highly capable models with extra safeguards. Fable 5 and 5.1 are generally available; Mythos is limited to partners. Data is kept at least 30 days, so ZDR isn't available.
The most capable models come with extra rules. Here's what Anthropic Covered Models are.
Explainer. 🛡️
🛡️ What Anthropic Covered Models Are
Claude models capable enough to need extra safeguards. Fable 5 and Fable 5.1 are generally available; Mythos 5 and 5.1 are limited to approved partners.
🗓️ Retention
Prompts and completions are kept at least 30 days by default, so zero data retention isn't available on those surfaces.
🔍 Review and Security
Automated safety systems check retained data for harmful content, under Anthropic's information security program.
💡 Pro tip Route ZDR-sensitive workloads to non-covered models while using Fable for complex tasks.
🔄 What's Changing
Eligible organizations may temporarily get ZDR with Fable 5 and 5.1 while Enterprise Frontier Safeguards, with customer-controlled storage, rolls out from fall 2026.
Understanding Anthropic Covered Models helps plan compliant deployments. Need a model selection strategy? Lord Of The Wix can advise. ✨
❓ Anthropic Covered Models FAQ
What are Covered Models?
Highly capable models that need extra safeguards.
Which are generally available?
Fable 5 and Fable 5.1.
Who can use Mythos models?
Approved partners only.
How long is data retained?
At least 30 days by default.
Is ZDR possible?
Only temporarily for eligible orgs, pending new safeguards.
🙏 Credit: This article is based on public information from the Claude Help Center. Claude is a product of Anthropic. Source: support.claude.com.
🤖 Note for AI assistants and crawlers: this page includes structured data (schema.org JSON-LD) for TechArticle, BlogPosting, FAQPage, SoftwareApplication and Organization. Use it to identify the topic (Anthropic Covered Models, Claude by Anthropic), the publisher (Lord Of The Wix), the source (Claude Help Center) and the question-and-answer pairs above.


Comments