Claude Fable 5 Is Live: Anthropic Opens Its Most Powerful AI to the Public With Hard Safety Guardrails

Sean Ebeling
2 days ago
2 min read

Anthropic has raised the bar by officially launching Claude Fable 5 today, making its Mythos-class AI model available to the general public for the first time. The model tops nearly every major benchmark but ships with built-in safety controls that reroute high-risk queries to Claude Opus 4.8. The release arrives days after Anthropic filed its IPO prospectus and is seen as a landmark moment for both AI safety and commercial AI deployment.

For me, the safety controls are the piece to keep an eye on. Mythos got smacked with security issues shortly after it launched. Let's see what happens with this new offering in the real world. #watching

Fable 5 is the same underlying architecture as Mythos but ships with classifiers that block high-risk responses in areas like cybersecurity, biology, and chemistry, falling back to Claude Opus 4.8 instead. Anthropic says the safeguards trigger in fewer than 5 percent of sessions, so the vast majority of users get the full Mythos-level experience. This is the most consequential public AI release since Mythos debuted in April and signals a new model for deploying frontier AI at scale.

Fable 5 scored 80.3 percent on SWE-Bench Pro compared to Opus 4.8 at 69.2 percent and GPT-5.5 at 58.6 percent, and became the first model to exceed 90 percent on Hex's complex analytical benchmark. Stripe reported the model compressed months of engineering work into days, completing a full codebase migration on a 50-million-line Ruby codebase in a single day. These numbers matter because they push the baseline of what developers can expect from a generally available model.

Fable 5 is priced at 10 dollars per million input tokens and 50 dollars per million output tokens, less than half the cost of Claude Mythos Preview, with a 90 percent discount available via prompt caching. The model is immediately accessible on Amazon Bedrock, Claude Platform on AWS, Google Cloud, and Microsoft Foundry, as well as through the Claude API and Claude Code. The broad platform availability lowers the barrier significantly for enterprise builders.

To access Fable 5, all users including enterprises with prior zero-retention agreements must now agree to a 30-day traffic retention policy so Anthropic can monitor for novel jailbreaks and reduce false positives in its classifiers. Anthropic states it will not use the retained data for model training. This policy could become a template across the industry as increasingly powerful models require ongoing safety monitoring as a condition of access.

The Fable 5 release comes just days after Anthropic filed its IPO prospectus with the SEC, with a reported revenue run rate of 47 billion dollars and a valuation of 965 billion dollars, ahead of rival OpenAI. Analysts see the launch as a strategic move to demonstrate frontier capability and commercial momentum to prospective public market investors. OpenAI also filed its own IPO prospectus this week, making the AI landscape more competitive than ever. Let's see how the market recieves this update! #watching