Claude Platform Updates: New Models, Tools, and AWS Integration
Anthropic has released significant updates to its Claude Platform, including new models like Claude Fable 5 and Claude Opus 4.8, which offer larger context windows and improved capabilities. Enhancements to tools like code execution and web search, alongside new features like mid-conversation system messages and improved refusal handling, provide developers with more control and efficiency. The platform also expands its availability with Claude Platform on AWS and self-hosted sandbox options, broadening access and integration for enterprise users.
- →Retirement of Claude Sonnet 4 and Opus 4 models
- →Code execution tool supports new version with execution time limit
- →Web search and fetch tools enhance response handling
- →Claude Platform on AWS gains new API endpoint for work listing
- →Launch of Claude Fable 5 and Claude Mythos 5 models
Deprecations (2) ›
- 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. All requests to these models will now result in an error. Users are recommended to upgrade to Claude Sonnet 4.6 and Claude Opus 4.8, respectively. Researchers can request continued access via the External Researcher Access Program.
- Deprecation of Claude Opus 4.1 model announced
The Claude Opus 4.1 model (claude-opus-4-1-20250805) has been deprecated, with its retirement on the Claude API scheduled for August 5, 2026. Users are advised to migrate to Claude Opus 4.8.
Features (6) ›
- Code execution tool supports new version with execution time limit
The code execution tool now supports code_execution_20260521, which explicitly states the 90-second per-cell execution time limit in its description. This allows Claude to better budget for long-running cells without requiring a beta header.
- Web search and fetch tools enhance response handling
The web search and web fetch tools now support new versions (web_search_20260318 and web_fetch_20260318) that include a 'response_inclusion' parameter. This parameter allows agents to drop consumed result blocks from the API response, which is useful for agentic workflows and does not require a beta header.
- Claude Platform on AWS gains new API endpoint for work listing
The GET /v1/environments/{id}/work endpoint, used to list pending work for self-hosted sandboxes, is now available on Claude Platform on AWS. Authorization for this action is managed through specific IAM actions for Claude Platform on AWS.
- Launch of Claude Fable 5 and Claude Mythos 5 models
Anthropic has 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. New tokenization, safety classifiers with refusal handling, and an opt-in fallback parameter are included.
- Advisor tool now supports max_tokens parameter
The advisor tool now includes a max_tokens parameter to limit the output per call, which can reduce latency and token costs for workloads not requiring full responses. This is configured via 'tools[].max_tokens' on the advisor tool definition.
- No billing for refused requests without output on Claude API
On the Claude API, users are no longer billed for requests that result in 'stop_reason: "refusal"' without any output being generated by Claude. This change aids in managing costs associated with refused requests.
https://platform.claude.com/docs/en/release-notes/overview