Claude Fable 5 powers the next wave of autonomous AI agents
The latest announcement shows that Claude Fable 5 is now available in Microsoft Foundry, giving Azure customers access to Anthropic’s newest frontier model that powers autonomous agents in GitHub Copilot and the Foundry Agent Service [source].
According to the model catalog, the system supports a context window of up to 1 million tokens and can generate up to 128K output tokens, enabling long-running reasoning and code synthesis [source].
Key capabilities include multi-stage planning, self-verification and visual reasoning for PDFs, charts and diagrams, which expands use cases for finance, legal and architecture teams [source].
Enterprises can integrate the model through GitHub Copilot or the Foundry Agent Service, and they can scale it with Azure governance, security and operational controls [source].
Getting started involves a few practical steps: ensure your Azure subscription is linked to Foundry, authenticate with Microsoft Entra ID using the DefaultAzureCredential library, and install the Anthropic SDK to call the Messages API. Sample code snippets for JavaScript and Python illustrate how to set up the client, set environment variables, and run a basic request that returns three recommendations [source].
Consider the broader implications: the model shifts AI workloads toward autonomous agents that can self-direct across multiple turns, which lowers operational overhead but also raises new governance responsibilities around safety filters, quota management and responsible AI compliance [source].