Anthropic Unveils Claude Fable 5: Powerhouse AI Model with Guardrails

Anthropic is taking a significant step in making cutting-edge artificial intelligence (AI) more accessible to the public with the launch of Claude Fable 5, the first publicly available version of its Mythos model. This powerful AI comes with robust guardrails designed to ensure safe and responsible use.

Fable 5 is optimized for tasks like software engineering, knowledge work, and vision. However, it includes strict safety limits that prevent responses in high-risk areas such as cybersecurity, biology, chemistry, and distillation. In these sensitive domains, the model falls back to Claude Opus 4.8.

Initially launched as a preview in April, Mythos was limited to select partners due to security concerns. Last week, Anthropic expanded access to hundreds of organizations across 15 countries, focusing on those managing critical infrastructure. Now, anyone can access Fable 5 through Anthropic’s Claude API and consumption-based Enterprise plans.

Access will roll out in stages: Through June 22, Fable 5 is included in Pro, Max, Team, and seat-based Enterprise plans at no extra cost. On June 23, it will be removed from these plans, requiring usage credits going forward. Anthropic aims to restore it as a standard subscription feature as soon as possible.

Anthropic has also deployed a new version of Mythos called Mythos 5 to approved organizations. This move comes as the company prepares for its public market debut alongside OpenAI and SpaceX, following its call for coordinated brakes on AI development.

The launch of Fable 5 is accompanied by stringent data retention policies. Anthropic will require a 30-day retention on all traffic, even if enterprises previously had zero-retention agreements. The company emphasizes that it won’t use the data for training but will leverage it to defend against complex and novel attacks.

Despite its powerful capabilities, Fable’s responses are not always generated by itself. In high-risk areas, it defers to Opus 4.8. However, early data shows that at least 95% of Fable sessions run entirely on the model's own responses. Third-party testing has also highlighted Fable's strengths in complex tasks and tool-calling.

Anthropic charges $10 per million input tokens and $50 per million output tokens for both Fable 5 and Mythos 5, double the price of Opus 4.8. This pricing could act as a deterrent for widespread use, especially given that many enterprises are growing critical about AI costs.

Despite the cost, some organizations see value in the advanced reasoning skills offered by Fable 5. For instance, shopping rewards platform Rakuten noted that Fable’s ability to reflect on and validate its own work makes it worth the price for highly autonomous operations.