/v1/messages when your client already speaks the Anthropic Messages format.
Tool use
Declare tools withtools (each has name, description, input_schema) and
control selection with tool_choice. Multi-turn tool_use / tool_result
content blocks are supported.
Streaming
Setstream: true to receive Anthropic server-sent events:
message_start, content_block_start, content_block_delta,
content_block_stop, message_delta, message_stop.
Parameters
Supported:model, messages, max_tokens (required), system,
temperature, top_p, top_k, stop_sequences, tools, tool_choice,
and stream.
Reasoning effort
Use nativeoutput_config.effort for supported Claude models. Supported values depend on the version; see model-specific reasoning controls. Native thinking is independent: effort does not enable thinking or select a token budget. If needed, configure thinking separately according to the model’s requirements.
max_completion_tokens, max_output_tokens, and the native equivalents include reasoning and final-answer tokens. Effort is not an output-token limit. A request can still exhaust its budget (finish_reason: "length" in Chat Completions) and return an incomplete or empty final answer.