← Back
Anthropic
Claude API adds model capability fields and extended thinking display controls

Models API Enhancement

The Claude Models API now includes model capability fields, enabling developers to programmatically discover what each model supports. The GET /v1/models and GET /v1/models/{model_id} endpoints now return:

  • max_input_tokens: Maximum input token capacity
  • max_tokens: Maximum total tokens supported
  • capabilities object: Feature support details

This allows applications to dynamically adapt to model capabilities without hardcoding version-specific assumptions.

Extended Thinking Display Controls

A new display field for extended thinking lets developers control whether thinking content appears in API responses. Setting thinking.display: "omitted" returns thinking blocks with an empty thinking field while preserving the signature field for multi-turn conversation continuity.

Key benefits:

  • Faster streaming by omitting verbose thinking content from responses
  • Maintains conversation context across turns via signature preservation
  • No billing changes—you pay the same regardless of display setting

Developers using extended thinking should review the Controlling thinking display documentation to optimize their implementations.