Claude Platform Updates: New Models, Rate Limits, and Tooling
Anthropic has released significant updates to its Claude Platform, including new models like Claude Fable 5 and Claude Opus 4.8 with larger context windows and enhanced capabilities. Rate limits have been raised across the API, and several tools have been updated with new features and versions. The updates also include deprecations and retirements of older models, alongside new SDK support and platform features for managed agents and AWS integration.
- →Launch of Claude Fable 5 and Claude Mythos 5
- →General Availability of Claude Opus 4.8
- →Fast Mode Deprecation for Claude Opus 4.7
- →Increased API Rate Limits and Consolidated Tiers
- →SDK Support for Code Execution Tool
Deprecations (1) ›
- Fast Mode Deprecation for Claude Opus 4.7
Fast mode for Claude Opus 4.7 is deprecated, with removal scheduled for July 24, 2026, requiring migration to fast mode for Claude Opus 4.8. Additionally, Claude Opus 4.1 is deprecated with retirement on August 5, 2026.
Features (2) ›
- Launch of Claude Fable 5 and Claude Mythos 5
Anthropic introduced Claude Fable 5 as its most capable widely released model and Claude Mythos 5 for specific participants. Both models offer a 1M token context window, 128k max output tokens, and always-on adaptive thinking, with Fable 5 also introducing a 'refusal' stop_reason for safety classifier declines.
- General Availability of Claude Opus 4.8
Claude Opus 4.8 is now generally available, featuring a 1M token context window on most platforms, 128k max output tokens, and support for mid-conversation system messages to preserve prompt cache hits when instructions change.
Enhancements (3) ›
- Increased API Rate Limits and Consolidated Tiers
Rate limits for Claude Sonnet and Haiku have been aligned with Claude Opus across all usage tiers, which have been consolidated into three: Start, Build, and Scale, with no organization experiencing reduced limits.
- SDK Support for Code Execution Tool
Python, TypeScript, Go, Java, Ruby, PHP, and C# SDKs now support code_execution_20260120, enabling REPL state persistence and programmatic tool calling with models like Claude Fable 5, Mythos 5, Opus 4.5+, and Sonnet 4.5+.
- Tool Updates for Code Execution and Web Search
The code execution tool now supports version code_execution_20260521 to disclose per-cell execution time limits, while web search and fetch tools support versions 20260318, adding a response_inclusion parameter to manage consumed result blocks.
https://platform.claude.com/docs/en/release-notes/overview