Source: aws.amazon.com/blogs/machine-learning/introducing-claude-fable-5-2-on-aws
Anthropic's Claude Fable 5.2 and Claude Mythos 5.2 are now available on Amazon Bedrock and Claude Platform on AWS. Anthropic describes Fable 5.2 as its most capable generally available model for technical work, including coding and science, and says it can run unattended for hours to days. Mythos 5.2 is available only through Anthropic's verification programs for cyber security and life sciences.
Fable 5.2 is designated a Covered Model, so by default its use requires data retention for up to 30 days and human safety review. On Bedrock, Amazon personnel perform that review within the AWS boundary, and developers set the aws_review data retention mode on their account or project before invoking the model. Enterprise Frontier Safeguards, built with AWS, is available now, and eligible customers can use zero data retention through June 30, 2027.
When a cyber security query is declined, the request can be retried on Opus 4.8, and a declined biology query can be retried on Opus 5. On Bedrock the retry is handled by the caller's code rather than an automatic switch. If the declined request hit the prompt cache, the refusal includes a fallback credit token that bills the retry as cache reads.
Why it matters if you build agents
Agents on Bedrock can now call Fable 5.2 for long, multi-step coding and research work, but refusals will not reroute on their own, so retry logic to a fallback model has to live in application code. Default retention and human review apply unless the aws_review mode is set or zero data retention is available, which matters for sensitive workloads. Sending the fallback credit token with retries avoids paying to write the same cache twice.
Get stories like this every weekday
One short email with the day's agent tooling news.