Anthropic Ships Fable 5.1 and Mythos 5.1
Anthropic released two new flagship models on September 1, 2026: Claude Fable 5.1, generally available through the Claude API, and Claude Mythos 5.1, restricted to vetted participants in Anthropic’s Project Glasswing trusted-access program. Both share the same underlying model and support a one-million-token context window with up to 128,000 tokens of output per response — Mythos simply runs with fewer safety classifiers for specialized cybersecurity and life-sciences work.
Fable 5.1 adds two notable features: the ability to adjust reasoning effort mid-conversation without starting a new session, and content provenance tracking that tags outputs so downstream systems can verify origin. Anthropic also says Claude Code users should see meaningfully fewer cybersecurity false positives, and that Fable 5.1 can now help surface software vulnerabilities (though not develop exploits for them). Cache-read pricing dropped sharply, cutting costs by roughly 25% for typical workloads and up to 45% for heavily agentic tasks.
Worth noting for EU-based teams: outputs now carry an invisible statistical watermark to comply with the EU AI Act, with a detection API rolling out in private preview.
Claudeforce: Anthropic and Salesforce Team Up
Anthropic and Salesforce announced an expanded partnership called Claudeforce, launching with a “Salesforce in Claude” plugin that bundles 37 prebuilt sales skills. The pitch: sellers and agents can reason over live revenue data, automate pipeline updates, and take governed actions directly inside Claude, with Salesforce’s workflows and business logic exposed as tools rather than static UI.
Free Access for K-12 Schools
Anthropic is now offering Claude for Teachers as a free Enterprise-tier product for schools and districts, including single sign-on, role-based access controls, and standards-aligned teaching tools. A pilot evaluation is underway with Detroit Public Schools, and qualifying districts can sign up through June 2027 for a year of free access.
Usage Limits Are Shifting Again
Developers relying on Claude Code should note that a temporary 50% boost to weekly usage limits is winding down, with a permanent 25% increase replacing it — a net cut from the promotional period’s ceiling. Teams budgeting Claude Code usage into their workflows should re-check their weekly caps rather than assume last month’s numbers still hold.
Why This Matters
The throughline across this week’s announcements is Anthropic pushing Claude deeper into regulated, high-stakes, and institutional territory — enterprise CRM workflows, K-12 classrooms, and specialized biosecurity/cybersecurity research access — while simultaneously tightening the screws on self-serve usage limits for individual developers. Worth watching whether that trade-off (more enterprise trust infrastructure, less generous default access) becomes a pattern with future releases.
Sources: Anthropic, Salesforce, MacRumors, TheNextWeb, DevOps.com — verified September 1–2, 2026.