Get the full response object from the last Interactions API call. Returns: Optional[Any]: The full interaction r

Get the most recent turn or None if history is empty.

Get or initialize the embedding model.

Get the Gemini client.

Get the OpenRouter client.

Get the model name (Ollama doesn't use a client object).

Get the ColBERT model.

Get the Mistral client.

Get the OpenAI client configured for Modular.

Get model by name (for backward compatibility)

Get model by name (for backward compatibility)

Get model by name (for backward compatibility)

Get model by name (for backward compatibility)

Get model by name (for backward compatibility)

Get model by name (for backward compatibility)

Get model by name (for backward compatibility)

Get a list of URLs from Parallel Search API. Args: objective: Natural language objective describing what you want to find

Get available model providers.

Compute log probabilities for each token in a sequence by performing a single forward pass through the model. Args:

Compute log probabilities using the Together completions API.

Retrieve the collected measurements. :return: List of tuples (timestamp, measurement dict)

Return performance statistics from the last request.

Get list of valid task types for embeddings.

Get the thinking content from the last model response that supports reasoning. Returns: str: The thinking conten

Get the total time the monitor has been running.

Get the most recent n turns. Args: n: Number of recent turns to retrieve, or None for all turns

Get URL context metadata from the last response. Returns: Optional[Dict[str, Any]]: URL context metadata if avai

Handle A2A JSON-RPC requests.

Handle WebSocket connection.

Handle WebSocket disconnection.

Check A2A server health.

Health check endpoint with detailed status.

Health check endpoint.

Health check endpoint.

Main page.

Display a list of jobs with toggleable details.

List all File Search stores. Returns: List[Dict[str, Any]]: List of file search store information

List uploaded files. Example: files = client.list_files(limit=20) for f in files: print(f"Fil

List available models from the remote OpenAI client. Returns: JSON response containing the list of available models

List available models from the NousResearch API. Returns: Dict containing the models data from the NousResearch

List available models from the MiniMax API. Returns: Dict containing the models data with structure: {

List available models from the Vercel AI Gateway. Returns a list of all models available through the gateway, including models from a

List available models from the Ollama Turbo API. Returns: Dict containing the models data

List available models from the Cohere API. Returns: Dict containing the models data from the Cohere API response

List available models from the Osaurus server. Calls the GET /models endpoint (OpenAI-compatible models list). Suppo

Load an audio file from a file path.

Load an audio file from a file path.

Load comparison results from a JSON file.

Explicitly load a model into memory.

Manage API keys for different providers.

Callback function to display intermediate MCP outputs

Stream messages from both local and remote models with real-time updates.

Show messages for both Minion and Minions protocols, labeling the local vs remote model clearly.

Multiply two numbers and return the result.

Chat with Nemotron models with explicit reasoning control. Args: messages: List of message dictionaries with 'ro

Return the number of tokens used by a list of messages. Source: https://github.com/openai/openai-cookbook/blob/main/examples/How_to_count_tokens_w

Handle OAuth2 token requests.

Parse the generated story text into structured chapters

Args: context (List[str]): A list of documents. Assume each document is greater >100k tokens. Each document can be further ch

Register a new node with the network coordinator.

Remove a file ID from chat attachments (does not delete the file). Args: file_id: The ID of the file to remove from atta

Remove a File Search store from being used in generations. Args: store_name: Name of the store to remove (format

Remove a node from the network.

Rerank documents based on relevance to a query using Cohere's rerank API. Requires native Cohere SDK. Args:

Use Perplexity's Agentic Research API for multi-provider access with web search. See: https://docs.perplexity.ai/docs/agenti

Use the OpenAI-compatible Responses API (requires Ollama v0.13.3+). Args: input: The input text/prompt

Use Exa's Responses API for single-turn research tasks. Example: client = ExaClient(model_name="exa-research