Anthropic made its Mythos-class AI model Claude Fable 5 publicly available on June 9, 2026, offering India’s broader public controlled access while enforcing a strict 30-day data retention policy to mitigate risks.
- Claude Fable 5 open to Indian public with 30-day data retention.
- High-risk Mythos 5 model remains restricted to vetted cybersecurity partners.
- Pricing set at $10 per million input tokens and $50 per million output tokens.
What happened
Anthropic released two Mythos-class AI models on June 9, 2026: Claude Fable 5, now accessible publicly, and Claude Mythos 5, which remains restricted to vetted cybersecurity partners. Both models operate on the same architecture and pricing scheme but differ in access and control policies.
A key feature of both models is a mandatory 30-day data retention requirement, diverging from previous zero-retention agreements. This is intended to enable Anthropic to monitor usage and defend against new types of attacks, even on third-party platforms. Fable 5 is deployed with an additional layer of AI-based classifiers that screen prompts, providing a safer, guard-railed experience for wider public users in India and other countries.
Why it matters
Claude Fable 5 represents the most capable AI model Anthropic has provided to the public, making advanced AI capabilities accessible beyond a limited group of organizations. However, the mandated data retention signals a cautious approach that prioritizes security and monitoring over user data privacy, which raises concerns for Indian enterprises and developers relying on API integrations.
The controlled release acknowledges the dual-edged nature of AI’s defensive and offensive potential. As analyst Nikhil Pahwa noted, expanding access without compressing defense timelines increases systemic risk. Anthropic’s retention and classifier safeguards aim to mitigate these risks but also indicate an industry trend toward embedding stricter surveillance with access to cutting-edge AI.
What to watch next
Anthropic plans to phase subscription models, initially including Fable 5 access under Pro, Max, Team, and Enterprise plans until June 22, 2026, before moving to a usage credit system. Observers should watch how these changes impact adoption among Indian developers and enterprises, especially regarding cost and data privacy trade-offs.
Additionally, the rollout of Anthropic’s trusted access program, which so far limits Mythos 5 to a small, undisclosed group including Indian government and private entities, will be key to watch. Monitoring how Anthropic balances broader access with security guardrails and whether challenges to safeguards emerge will be important for understanding the evolving risk landscape in India and beyond.