orbiter.models.openai

OpenAI LLM provider implementation. Wraps the openai SDK to implement ModelProvider.complete() and ModelProvider.stream() with normalized response types.

Module Path

from orbiter.models.openai import OpenAIProvider

Auto-Registration

On import, OpenAIProvider is registered in model_registry under the name "openai":

model_registry.register("openai", OpenAIProvider)

OpenAIProvider

Wraps the openai.AsyncOpenAI client for chat completions.

Inherits: ModelProvider

Constructor

OpenAIProvider(config: ModelConfig)

The constructor creates an AsyncOpenAI client using:

• api_key from config.api_key (falls back to "dummy" if None)
• base_url from config.base_url
• max_retries from config.max_retries
• timeout from config.timeout

Methods

complete()

async def complete(
    self,
    messages: list[Message],
    *,
    tools: list[dict[str, Any]] | None = None,
    temperature: float | None = None,
    max_tokens: int | None = None,
) -> ModelResponse

Send a chat completion request to OpenAI.

Returns: ModelResponse

Raises: ModelError if the API call fails.

stream()

async def stream(
    self,
    messages: list[Message],
    *,
    tools: list[dict[str, Any]] | None = None,
    temperature: float | None = None,
    max_tokens: int | None = None,
) -> AsyncIterator[StreamChunk]

Stream a chat completion from OpenAI. Automatically sets stream=True and stream_options={"include_usage": True} to get token usage on the final chunk.

Yields: StreamChunk

Raises: ModelError if the API call fails.

Message Conversion

Orbiter messages are converted to OpenAI format internally:

Finish Reason Mapping

Reasoning Models

For o1/o3 reasoning models, the reasoning_content field in ModelResponse is populated from the response’s model_extra["reasoning_content"] if present.

Example

import asyncio
from orbiter.models import get_provider
from orbiter.types import SystemMessage, UserMessage

async def main():
    provider = get_provider("openai:gpt-4o", api_key="sk-...")

    response = await provider.complete(
        [
            SystemMessage(content="You are a helpful assistant."),
            UserMessage(content="What is the capital of France?"),
        ],
        temperature=0.0,
        max_tokens=100,
    )

    print(response.content)  # "The capital of France is Paris."
    print(response.usage)    # Usage(input_tokens=25, output_tokens=8, total_tokens=33)

asyncio.run(main())

Compatible Services

Since OpenAIProvider wraps the standard OpenAI SDK, it works with any OpenAI-compatible API by setting base_url:

• OpenAI (default)
• Azure OpenAI (set base_url to your Azure endpoint)
• vLLM (base_url="http://localhost:8000/v1")
• Ollama (base_url="http://localhost:11434/v1")
• LiteLLM proxy