← Back
Anthropic
Anthropic launches automatic prompt caching and Claude Sonnet 4.6; retires earlier models
Claude · releasefeatureapimodeldeprecation · platform.claude.com ↗

New Features

Automatic Prompt Caching is now available on the Claude API and Azure AI Foundry (preview). Developers can add a single cache_control field to their request body and the system automatically caches the last cacheable block, advancing the cache point as conversations grow. This eliminates manual breakpoint management while working alongside existing block-level cache control for fine-grained optimization.

Claude Sonnet 4.6 is Anthropic's latest balanced model, combining speed and intelligence for everyday tasks. It delivers improved agentic search performance while consuming fewer tokens. The model supports extended thinking and a 1M token context window (beta).

Model Retirements & Deprecations

  • Retired immediately: Claude Sonnet 3.7 (claude-3-7-sonnet-20250219) and Claude Haiku 3.5 (claude-3-5-haiku-20241022) — all requests now return errors. Users should upgrade to Claude Sonnet 4.6 and Claude Haiku 4.5 respectively.
  • Deprecated with sunset date: Claude Haiku 3 (claude-3-haiku-20240307) will be retired on April 19, 2026. Migration to Claude Haiku 4.5 is recommended.
  • Researchers can request ongoing access through the External Researcher Access Program.

Developer Actions Required

Developers using the retired models (Sonnet 3.7, Haiku 3.5) must immediately update their API calls to use the recommended replacement models. For Haiku 3, migration should be planned before April 19, 2026.