---
type: "post"
id: "introducing-claude-fable-5-2-on-aws-4ac017"
slug: "introducing-claude-fable-5-2-on-aws-4ac017"
title: "Introducing Claude Fable 5.2 on AWS"
description: "AWS says Claude Fable 5.2 and Claude Mythos 5.2 are available through Amazon Bedrock and Claude Platform on AWS."
retrieval_nugget: "AWS says Claude Fable 5.2 and Claude Mythos 5.2 are available through Amazon Bedrock and Claude Platform on AWS. Fable is generally available for technical work; Mythos remains restricted to Anthropic verification programs. For builders, the important details are deployment access, retention requirements and application-managed fallback behavior."
published_at: "2026-11-10"
updated_at: "2026-11-10"
record_date: "2026-11-10"
date_kind: "published_at"
topics: ["claude","anthropic"]
entities: ["Amazon Web Services","Amazon Bedrock","Claude Fable 5.2","Claude Mythos 5.2","Claude Platform on AWS","Anthropic"]
related_patterns: ["agent-economics-moves-to-completed-work"]
novelty_label: "new_fact"
what_changed: "Claude Fable 5.2 and Mythos 5.2 are now available on Amazon Bedrock."
prior_refs: []
source_url: "https://aws.amazon.com/blogs/machine-learning/introducing-claude-fable-5-2-on-aws"
source_urls: ["https://aws.amazon.com/blogs/machine-learning/introducing-claude-fable-5-2-on-aws"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:introducing-claude-fable-5-2-on-aws-4ac017"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/introducing-claude-fable-5-2-on-aws-4ac017/","markdown":"https://newruntime.com/posts/introducing-claude-fable-5-2-on-aws-4ac017.md","json":"https://newruntime.com/posts/introducing-claude-fable-5-2-on-aws-4ac017.json"}
markdown_url: "https://newruntime.com/posts/introducing-claude-fable-5-2-on-aws-4ac017.md"
---

# Introducing Claude Fable 5.2 on AWS

## Retrieval answer

AWS says Claude Fable 5.2 and Claude Mythos 5.2 are available through Amazon Bedrock and Claude Platform on AWS. Fable is generally available for technical work; Mythos remains restricted to Anthropic verification programs. For builders, the important details are deployment access, retention requirements and application-managed fallback behavior.

## Summary
AWS says Claude Fable 5.2 and Claude Mythos 5.2 are available through Amazon Bedrock and Claude Platform on AWS. Fable is generally available for technical work; Mythos remains restricted to Anthropic verification programs. For builders, the important details are deployment access, retention requirements and application-managed fallback behavior.

## What happened
AWS announced availability of Claude Fable 5.2 and Claude Mythos 5.2 on Amazon Bedrock and Claude Platform on AWS. It describes Fable 5.2 as Anthropic’s most capable generally available model for technical work, including coding and the sciences. Mythos 5.2 is positioned for advanced cybersecurity and life-sciences work and is accessible only through Anthropic’s verification programs.

The AWS post says Fable 5.2 is intended for ambiguous, multi-step tasks, can iterate and check its output, and can run unattended for hours to days. Listed uses include coding across large multi-repository codebases, long-running agents coordinating subagents, scientific research and formal proofs, quantitative financial analysis, and code-generated designs or simulations for 3D, hardware and motion. Mythos use cases are security engineering, red teaming, penetration testing, biomolecular design and research synthesis. AWS also relays Anthropic alignment testing that found Mythos 5.2 hallucinated less and completed tasks more reliably than Opus 5.5, Sonnet 5.5 and Mythos 5.1; AWS says Anthropic expects Fable 5.2 to behave similarly.

There are operational conditions. Anthropic designates Fable 5.2 a Covered Model, and AWS says default use requires retention for up to 30 days and human safety review. On Bedrock, Amazon personnel perform that review within the AWS boundary. Customers must set the `aws_review` retention mode on an account or project before invoking the model. AWS describes Enterprise Frontier Safeguards as combining zero data retention with misuse safeguards, with data stored in customer-controlled cloud infrastructure. Until that is available, eligible customers can use Fable 5.2 with zero data retention for internal use through at least June 30, 2027.

AWS says both models are available today on Bedrock through US Geo CRIS and Global CRIS inference profiles, and through Claude Platform on AWS in every region where that service is offered. Builders can select Fable 5.2 in the Bedrock console or call it through the Anthropic Messages API, Invoke or Converse APIs. AWS’s example uses the model ID `global.anthropic.claude-fable-5-2`. If a cybersecurity request is declined, AWS says it can be retried on Opus 4.8; a declined biology request can be retried on Opus 5. On Bedrock, the application must make that retry rather than relying on an automatic switch. A fallback credit token can prevent paying to write the same prompt cache twice.

## What the archive adds
[New Runtime's earlier coverage](https://newruntime.com/posts/introducing-claude-fable-5-2-and-claude-mythos-6b1c41/) on November 9 covered Fable 5.2 formalizing Fermat’s Last Theorem in 30 hours. This is a new access and deployment fact about the same models, not a restatement of that research result.

## Trend signal
This supports [agent economics moves to completed work](https://newruntime.com/patterns/agent-economics-moves-to-completed-work/) only weakly: AWS emphasizes long-running work and multi-step task execution, but gives no task-completion cost or independent success-rate comparison. The announcement makes workload scope visible without yet establishing the economics of completed work.

## Why now
AWS presents the release as available today and supplies a migration path for teams already running Claude models on Bedrock, including a Claude Code skill that updates model references, thinking settings and parameters. The post does not explain why the release is timed for this date.

## Why it matters
This is relevant less as a new model announcement than as a deployment decision. Teams already operating in AWS can test Fable 5.2 through existing Bedrock interfaces rather than building a separate integration, but they need to account for model-specific retention and human review before production use. The 30-day default retention condition may conflict with internal policies, while the described zero-retention option is limited to eligible customers and internal use during the stated interim period.

The fallback behavior also belongs in application design: an eligible refusal does not automatically route to another model on Bedrock. Teams that need continuity should explicitly implement the retry and pass the fallback credit token when applicable. Mythos 5.2 should not be treated as a generally available substitute for Fable; its intended domains and access remain bounded by Anthropic’s verification programs.

## Limits and uncertainty
AWS does not give benchmark scores, pricing figures or a quantified comparison showing how often Fable 5.2 completes the workloads it lists. Its summary of alignment testing is not an independent evaluation, and the expectation that Fable behaves similarly is not itself a reported result for Fable. The post points readers to Bedrock pricing and regional documentation rather than enumerating costs or every supported region. Access eligibility for Mythos and zero-retention use depends on programs or customer status not detailed in the announcement.

## Skill to build now
Run a controlled migration review before changing production traffic. Use the documented model ID, inventory each existing model reference, and confirm whether manual thinking configuration, deprecated headers, or prompt-cache behavior needs changing. Then test the refusal path as an application workflow: record which requests are eligible for retry, where the fallback credit token is passed, and whether policy permits the selected retention mode. This turns a model swap into an auditable deployment change rather than a string replacement.

## Reader takeaway
For AWS-based teams, Fable 5.2 is now an available option for demanding technical work, but availability alone does not settle suitability. Check the regional and pricing details, confirm retention eligibility, and implement refusal fallback in your own orchestration before placing it into a production workflow.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Claude](https://newruntime.com/topics/claude.md): Explore the claude topic hub.
- topic hub: [Anthropic](https://newruntime.com/topics/anthropic.md): Explore the anthropic topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)