ReqLLM.Providers.Anthropic.Response (ReqLLM v1.0.0)

View Source

Anthropic-specific response decoding for the Messages API format.

Handles decoding Anthropic Messages API responses to ReqLLM structures.

Anthropic Response Format

%{
  "id" => "msg_01XFDUDYJgAACzvnptvVoYEL",
  "type" => "message",
  "role" => "assistant",
  "model" => "claude-3-5-sonnet-20241022",
  "content" => [
    %{"type" => "text", "text" => "Hello! How can I help you today?"}
  ],
  "stop_reason" => "stop",
  "stop_sequence" => nil,
  "usage" => %{
    "input_tokens" => 10,
    "output_tokens" => 20
  }
}

Streaming Format

Anthropic uses Server-Sent Events (SSE) with different event types:

  • message_start: Initial message metadata
  • content_block_start: Start of content block
  • content_block_delta: Incremental content
  • content_block_stop: End of content block
  • message_delta: Final message updates
  • message_stop: End of message

Summary

Functions

decode_response(data, model)

Decode Anthropic response data to ReqLLM.Response.

decode_stream_event(arg1, model)

Decode Anthropic SSE event data into StreamChunks.

Functions

decode_response(data, model)

@spec decode_response(map(), ReqLLM.Model.t()) ::
  {:ok, ReqLLM.Response.t()} | {:error, term()}

Decode Anthropic response data to ReqLLM.Response.

decode_stream_event(arg1, model)

@spec decode_stream_event(map(), ReqLLM.Model.t()) :: [ReqLLM.StreamChunk.t()]

Decode Anthropic SSE event data into StreamChunks.