AWS adds Claude Fable 5.1 to Bedrock & AWS platform
Tue, 8th Sep 2026 (Today)
AWS has made Claude Fable 5.1 available on its cloud platform through Amazon Bedrock and Claude Platform on AWS.
The launch adds another advanced Anthropic model to AWS as cloud providers compete to host more artificial intelligence systems for coding, research and business tasks. Anthropic has classified Claude Fable 5.1 as a Covered Model, AWS said, bringing additional rules on data retention, safety review and access.
Under those rules, prompts and outputs can be retained for up to 30 days and reviewed by Amazon personnel for safety purposes. AWS said the arrangement uses a new aws_review data retention mode, while Amazon Bedrock does not share customer data with the model provider under the legacy provider_data_share mode.
Eligible customers will also be able to use Covered Models through Enterprise Frontier Safeguards, a setup developed with Anthropic to keep customer data inside a cloud environment under the customer's control. That addresses a central issue for large organisations weighing how to use advanced models while maintaining internal governance and compliance controls.
Claude Fable 5.1 is designed for longer tasks that can run across several applications and over extended periods. AWS said it can work across an entire software codebase, including feature development, code review and performance-related tasks.
Broader rollout
The model launch was part of a wider batch of AWS product updates spanning infrastructure, serverless computing, data systems and AI management tools. Together, the announcements show AWS continuing to add new services and refresh existing ones as demand grows for cloud systems that can support AI workloads alongside conventional applications.
Among the other releases, AWS put Amazon Linux 2027 into public preview. The operating system is the next version of Amazon Linux and builds on the baseline established by Amazon Linux 2023.
AWS said the new version runs on kernel 7.1 or later and is intended for cloud-native workloads on its platform. It is aimed at customers running web applications, databases, containerised microservices, AI and machine learning workloads, and large infrastructure estates.
In computing infrastructure, Amazon EC2 R9g and R9gd memory-optimised instances are now available, AWS said. The instances use AWS Graviton5 processors and target memory-intensive workloads on EC2.
According to AWS, the instances offer up to 25% better compute performance than Graviton4-based R8g and R8gd instances. They are also up to 30% faster for databases, up to 35% faster for web applications and up to 35% faster for machine learning tasks.
Serverless and data
AWS has also expanded Lambda SnapStart to container image functions. The feature had previously been limited to managed runtimes including Python, .NET and Java.
With the change, customers using container images can reduce application startup times from several seconds to sub-second levels, according to AWS. The company said this is relevant for latency-sensitive uses such as machine learning inference and interactive application programming interfaces.
Another update moved AWS Agent Registry into general availability. AWS described the service as a private catalog and discovery layer for agents, tools, skills, MCP servers and custom resources within an organisation.
The generally available version includes features introduced in preview, such as manual and URL-based record creation, approval workflows, semantic and keyword search, and AWS CloudTrail audit trails. AWS said the release also adds new features for larger organisations, though it did not detail them in the same summary.
On the data side, Amazon Redshift now supports Apache Iceberg v3 tables. Users can now read from and write to Apache Iceberg v3 tables in data lakes from Amazon Redshift.
AWS said the update adds support for default column values, row lineage and deletion vectors. The new format is supported on Graviton-based provisioned and serverless Redshift clusters.
Skills and positioning
Beyond product releases, AWS also outlined moves in training and industry positioning. It introduced AWS Certified AI Business Strategist, a certification aimed at professionals responsible for assessing, promoting and expanding AI initiatives in their organisations.
The certification is targeted at line-of-business leaders, sales professionals, consultants and programme managers. That points to AWS's effort to widen AI adoption beyond software engineers and data scientists to executives and commercial teams tasked with linking AI spending to business outcomes.
AWS also highlighted a security initiative focused on agentic AI workloads, referring to a framework produced with the SANS Institute on detection, response and architectural patterns for AI systems. The emphasis reflects growing concern across the sector over how autonomous or semi-autonomous software agents should be monitored and governed when deployed at scale.
The latest set of announcements places Claude Fable 5.1 within a broader AWS strategy that combines access to advanced foundation models with controls around data handling, infrastructure tuned for heavier AI use, and tools for managing more complex agent-based systems. It also shows AWS trying to address the operational questions that emerge once customers move from experimenting with AI models to using them in software projects, internal workflows and large data environments.
Amazon Bedrock and Claude Platform on AWS are the two routes for customers seeking access to Claude Fable 5.1, AWS said.