Models API Enhanced with Capability Metadata

The Models API now returns comprehensive model capability information. Calls to GET /v1/models and GET /v1/models/{model_id} now include:

• max_input_tokens: Maximum input context length for the model
• max_tokens: Maximum output token generation
• capabilities: An object describing supported features

This allows developers to programmatically discover and query model capabilities at runtime, enabling dynamic model selection and feature detection.

Extended Thinking Display Control

Extended thinking now supports a display field that allows you to omit thinking content from responses. Set thinking.display: "omitted" to:

• Receive thinking blocks with an empty thinking field
• Preserve the signature for multi-turn conversation continuity
• Enable faster response streaming
• Keep billing unchanged

This addresses use cases where internal reasoning is unnecessary for the client while maintaining conversation state across turns.