Model API

Model API

Public exports

agiwo.llm exports:

• LLMConfig
• Model
• StreamChunk
• ModelSpec
• ModelProvider
• create_model(...)
• create_model_from_dict(...)
• OpenAIModel
• OpenAIResponsesModel
• AnthropicModel
• BedrockAnthropicModel
• DeepseekModel
• NvidiaModel

Model

The shared base class for all providers.

class Model(ABC):
    config: LLMConfig

    async def arun_stream(
        self,
        messages: list[dict],
        tools: list[dict] | None = None,
    ) -> AsyncIterator[StreamChunk]: ...

    async def close(self) -> None: ...

StreamChunk

Each provider yields normalized streaming deltas:

Field	Purpose
content	Text delta
reasoning_content	Reasoning delta when supported
tool_calls	Tool call delta
usage	Token usage
finish_reason	Provider finish reason

Typical construction

from agiwo.llm import OpenAIModel

model = OpenAIModel(name="gpt-5.4")

from agiwo.llm import create_model_from_dict

model = create_model_from_dict(
    provider="openai-compatible",
    model_name="custom-model",
    params={
        "base_url": "https://api.example.com/v1",
        "api_key_env_name": "MY_API_KEY",
    },
)

Supported provider strings

• openai
• openai-response
• openai-compatible
• deepseek
• anthropic
• anthropic-compatible
• nvidia
• bedrock-anthropic