pi_llm.stream

stream

Top-level streaming API.

The four functions here are the primary interface for making LLM calls. Use stream_simple / complete_simple for most use cases. Use stream / complete when you need provider-specific options.

stream(model, context, options=None)

Stream a response using provider-specific options.

Parameters:

Name	Type	Description	Default
model	Model	The model to use.	required
context	Context	Conversation context (system prompt, messages, tools).	required
options	StreamOptions | None	Provider-specific streaming options.	None

Returns:

Type	Description
AssistantMessageEventStream	An async iterable event stream. Use async for event in stream
AssistantMessageEventStream	to consume events, or await stream.result() for the final message.

Raises:

Type	Description
ValueError	If no provider is registered for the model's API.

complete(model, context, options=None) async

Non-streaming convenience wrapper around stream().

Parameters:

Name	Type	Description	Default
model	Model	The model to use.	required
context	Context	Conversation context.	required
options	StreamOptions | None	Provider-specific streaming options.	None

Returns:

Type	Description
AssistantMessage	The completed assistant message.

stream_simple(model, context, options=None)

Stream a response with automatic reasoning support.

This is the recommended entry point for most use cases. It handles reasoning level mapping and option building automatically.

Parameters:

Name	Type	Description	Default
model	Model	The model to use.	required
context	Context	Conversation context (system prompt, messages, tools).	required
options	SimpleStreamOptions | None	Simplified options including reasoning level.	None

Returns:

Type	Description
AssistantMessageEventStream	An async iterable event stream.

Raises:

Type	Description
ValueError	If no provider is registered for the model's API.

complete_simple(model, context, options=None) async

Non-streaming convenience wrapper around stream_simple().

Parameters:

Name	Type	Description	Default
model	Model	The model to use.	required
context	Context	Conversation context.	required
options	SimpleStreamOptions | None	Simplified options including reasoning level.	None

Returns:

Type	Description
AssistantMessage	The completed assistant message.