Claude Platform Updates: New Models, Tools, and AWS Integration
Anthropic has released new models like Claude Fable 5 and Claude Opus 4.8, featuring expanded context windows and improved adaptive thinking. Updates to tools like the code execution and web search tools enhance agentic workflows. Key features include mid-conversation system messages and expanded Claude Managed Agents capabilities on AWS. These updates are relevant to developers and architects integrating Claude into applications.
- →Retirement of Claude Sonnet 4 and Opus 4 models
- →Code execution tool supports new version with time limit disclosure
- →Web tools add response inclusion parameter
- →Self-hosted sandbox work endpoint now on Claude Platform on AWS
- →Launch of Claude Fable 5 and Claude Mythos 5
Deprecations (1) ›
- Retirement of Claude Sonnet 4 and Opus 4 models
The Claude Sonnet 4 (claude-sonnet-4-20250514) and Claude Opus 4 (claude-opus-4-20250514) models have been retired. Requests to these models will now result in an error, and users are recommended to upgrade to Claude Sonnet 4.6 and Claude Opus 4.8 respectively.
Features (4) ›
- Code execution tool supports new version with time limit disclosure
The code execution tool now supports code_execution_20260521, which specifies a 90-second per-cell execution time limit. This allows Claude to better manage long-running cells without requiring a beta header.
- Web tools add response inclusion parameter
The web search and web fetch tools now support a response_inclusion parameter (web_search_20260318 and web_fetch_20260318). This parameter allows dropping consumed result blocks from the API response for agentic workflows, without requiring a beta header.
- Self-hosted sandbox work endpoint now on Claude Platform on AWS
The GET /v1/environments/{id}/work endpoint, which lists pending work for self-hosted sandboxes, is now available on Claude Platform on AWS. The GetEnvironment IAM action authorizes this functionality.
- Launch of Claude Fable 5 and Claude Mythos 5
Anthropic launched Claude Fable 5 (claude-fable-5), their most capable widely released model, and Claude Mythos 5 (claude-mythos-5) for Project Glasswing participants. Both models feature a 1M token context window, 128k max output tokens, and adaptive thinking. Claude Fable 5 introduces new safety classifier behavior with a stop_reason: "refusal" and an optional fallback parameter for rerunning refused requests.
Enhancements (3) ›
- New tokenizer increases token count for same text
Claude Fable 5 and Claude Mythos 5 utilize a new tokenizer introduced with Claude Opus 4.7. This results in approximately 30% more tokens for the same text compared to older models. A token counting API with model: "claude-fable-5" is available for accurate prompt measurement.
- Claude Managed Agents enhancements
Claude Managed Agents now support scheduled deployments for cron-based execution and vaults with environment variable credentials for secure secret injection into agent sandboxes.
- Session thread webhooks now include thread ID
The session.thread_* webhook events now provide a session_thread_id field, enabling identification of the multi-agent thread that triggered the event.
https://platform.claude.com/docs/en/release-notes/overview