ant_ai.core.response

ChatLLMResponse pydantic-model

Bases: BaseModel

Response from a chat-based LLM. It follows a OpenAI-like structure.

Show JSON schema:

{
  "$defs": {
    "Message": {
      "description": "Generic message used in a conversation",
      "properties": {
        "kind": {
          "const": "message",
          "default": "message",
          "title": "Kind",
          "type": "string"
        },
        "role": {
          "title": "Role",
          "type": "string"
        },
        "content": {
          "anyOf": [
            {
              "type": "string"
            },
            {
              "type": "null"
            }
          ],
          "default": null,
          "title": "Content"
        },
        "metadata": {
          "additionalProperties": true,
          "title": "Metadata",
          "type": "object"
        }
      },
      "required": [
        "role"
      ],
      "title": "Message",
      "type": "object"
    },
    "ToolCall": {
      "description": "Single tool call object inside assistant.tool_calls (OpenAI schema).",
      "properties": {
        "id": {
          "title": "Id",
          "type": "string"
        },
        "type": {
          "default": "function",
          "title": "Type",
          "type": "string"
        },
        "function": {
          "$ref": "#/$defs/ToolFunction"
        }
      },
      "required": [
        "id",
        "function"
      ],
      "title": "ToolCall",
      "type": "object"
    },
    "ToolFunction": {
      "description": "Inner function payload for a tool call (OpenAI schema).",
      "properties": {
        "name": {
          "title": "Name",
          "type": "string"
        },
        "arguments": {
          "title": "Arguments",
          "type": "string"
        }
      },
      "required": [
        "name",
        "arguments"
      ],
      "title": "ToolFunction",
      "type": "object"
    }
  },
  "description": "Response from a chat-based LLM. It follows a OpenAI-like structure.",
  "properties": {
    "message": {
      "$ref": "#/$defs/Message",
      "description": "The assistant message returned by the model."
    },
    "tool_calls": {
      "description": "Tool calls requested by the model. Empty when the model produced a text answer.",
      "items": {
        "$ref": "#/$defs/ToolCall"
      },
      "title": "Tool Calls",
      "type": "array"
    },
    "finish_reason": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "The reason the model stopped generating (e.g. 'stop', 'tool_calls').",
      "title": "Finish Reason"
    },
    "usage": {
      "anyOf": [
        {
          "additionalProperties": true,
          "type": "object"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "Token usage statistics reported by the model.",
      "title": "Usage"
    },
    "reasoning": {
      "anyOf": [
        {
          "type": "string"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "Reasoning/thinking content produced by the model, if supported.",
      "title": "Reasoning"
    }
  },
  "required": [
    "message"
  ],
  "title": "ChatLLMResponse",
  "type": "object"
}

Fields:

• message (Message)
• tool_calls (list[ToolCall])
• finish_reason (str | None)
• usage (dict | None)
• reasoning (str | None)

Source code in src/ant_ai/core/response.py

class ChatLLMResponse(BaseModel):
    """Response from a chat-based LLM. It follows a OpenAI-like structure."""

    message: Message = Field(
        description="The assistant message returned by the model.",
    )
    tool_calls: list[ToolCall] = Field(
        default_factory=list,
        description="Tool calls requested by the model. Empty when the model produced a text answer.",
    )
    finish_reason: str | None = Field(
        default=None,
        description="The reason the model stopped generating (e.g. 'stop', 'tool_calls').",
    )
    usage: dict | None = Field(
        default=None,
        description="Token usage statistics reported by the model.",
    )
    reasoning: str | None = Field(
        default=None,
        description="Reasoning/thinking content produced by the model, if supported.",
    )

message pydantic-field

message: Message

The assistant message returned by the model.

tool_calls pydantic-field

tool_calls: list[ToolCall]

Tool calls requested by the model. Empty when the model produced a text answer.

finish_reason pydantic-field

finish_reason: str | None = None

The reason the model stopped generating (e.g. 'stop', 'tool_calls').

usage pydantic-field

usage: dict | None = None

Token usage statistics reported by the model.

reasoning pydantic-field

reasoning: str | None = None

Reasoning/thinking content produced by the model, if supported.

ChatLLMStreamChunk pydantic-model

Bases: BaseModel

A chunk of a streaming response from a chat-based LLM.

Show JSON schema:

{
  "$defs": {
    "MessageChunk": {
      "description": "Represents *partial output* from an LLM during streaming.\n`delta` is the content's only the newly streamed text fragment,\nnot the whole message so far.",
      "properties": {
        "role": {
          "title": "Role",
          "type": "string"
        },
        "delta": {
          "title": "Delta",
          "type": "string"
        },
        "metadata": {
          "additionalProperties": true,
          "title": "Metadata",
          "type": "object"
        }
      },
      "required": [
        "role",
        "delta"
      ],
      "title": "MessageChunk",
      "type": "object"
    }
  },
  "description": "A chunk of a streaming response from a chat-based LLM.",
  "properties": {
    "delta": {
      "$ref": "#/$defs/MessageChunk",
      "description": "Newly streamed text fragment, not the accumulated response so far."
    },
    "tool_calls": {
      "anyOf": [
        {
          "additionalProperties": true,
          "type": "object"
        },
        {
          "type": "null"
        }
      ],
      "default": null,
      "description": "Partial tool call data streamed incrementally alongside the delta.",
      "title": "Tool Calls"
    }
  },
  "required": [
    "delta"
  ],
  "title": "ChatLLMStreamChunk",
  "type": "object"
}

Fields:

• delta (MessageChunk)
• tool_calls (dict | None)

Source code in src/ant_ai/core/response.py

class ChatLLMStreamChunk(BaseModel):
    """A chunk of a streaming response from a chat-based LLM."""

    delta: MessageChunk = Field(
        description="Newly streamed text fragment, not the accumulated response so far.",
    )
    tool_calls: dict | None = Field(
        default=None,
        description="Partial tool call data streamed incrementally alongside the delta.",
    )

delta pydantic-field

delta: MessageChunk

Newly streamed text fragment, not the accumulated response so far.

tool_calls pydantic-field

tool_calls: dict | None = None

Partial tool call data streamed incrementally alongside the delta.