Skip to main content

​
module  pixeltable.functions.openrouter

Pixeltable UDFs that wrap the OpenRouter API. OpenRouter provides a unified interface to multiple LLM providers. In order to use it, you must first sign up at https://openrouter.ai, create an API key, and configure it as described in the Working with OpenRouter tutorial.

​
udf  chat_completions()

chat_completions(
    messages: Json,
    *,
    model: String,
    model_kwargs: Json | None = None,
    tools: Json | None = None,
    tool_choice: Json | None = None,
    provider: Json | None = None,
    transforms: Json | None = None
) -> Json
Chat Completion API via OpenRouter. OpenRouter provides access to multiple LLM providers through a unified API. For additional details, see: https://openrouter.ai/docs Supported models can be found at: https://openrouter.ai/models Request throttling: Applies the rate limit set in the config (section openrouter, key rate_limit). If no rate limit is configured, uses a default of 600 RPM. Requirements:
  • pip install openai
Parameters:
  • messages (Json): A list of messages comprising the conversation so far.
  • model (String): ID of the model to use (e.g., 'anthropic/claude-3.5-sonnet', 'openai/gpt-4').
  • model_kwargs (Json | None): Additional OpenAI-compatible parameters.
  • tools (Json | None): List of tools available to the model.
  • tool_choice (Json | None): Controls which (if any) tool is called by the model.
  • provider (Json | None): OpenRouter-specific provider preferences (e.g., {'order': ['Anthropic', 'OpenAI']}).
  • transforms (Json | None): List of message transforms to apply (e.g., ['middle-out']).
Returns:
  • Json: A dictionary containing the response in OpenAI format.
Examples: Basic chat completion: 
messages = [{'role': 'user', 'content': tbl.prompt}]
tbl.add_computed_column(
    response=chat_completions(
        messages, model='anthropic/claude-3.5-sonnet'
    )
)
With provider routing: 
tbl.add_computed_column(
    response=chat_completions(
        messages,
        model='anthropic/claude-3.5-sonnet',
        provider={'require_parameters': True, 'order': ['Anthropic']},
    )
)
With transforms: 
tbl.add_computed_column(
    response=chat_completions(
        messages,
        model='openai/gpt-4',
        transforms=['middle-out'],  # Optimize for long contexts
    )
)