dataflow.llms.api

API endpoints for LLM model information and availability.

This module provides FastAPI endpoints to access and manage LLM model data stored in Supabase. It helps bridge the client application with the database while providing additional server-side logic.

Functions

get_capabilities()	Get all possible LLM capabilities.
get_model_by_id(model_id)	Get a specific model by ID with all related information.
get_models([provider])	Get all models from the database.
get_modes()	Get all possible LLM operation modes.
get_providers()	Get all providers from the database.
read_model(model_id)	Get a specific model by ID.
read_models([provider, capability])	Get all models with optional filtering.
read_providers()	Get all LLM providers.
recommended_models([task, vision, function_calling, ...])	Get recommended models based on capabilities and task requirements.

Module Contents

async dataflow.llms.api.get_capabilities()

Get all possible LLM capabilities.

dataflow.llms.api.get_model_by_id(model_id)

Get a specific model by ID with all related information.

Parameters:

model_id (str) – The model ID to look up

Returns:

Model data with capabilities and pricing, or None if not found

Return type:

dict[str, Any] | None

dataflow.llms.api.get_models(provider=None)

Get all models from the database.

Parameters:

provider (str | None) – Optional provider name to filter by

Returns:

List of models with their capabilities and pricing

Return type:

list[dict[str, Any]]

async dataflow.llms.api.get_modes()

Get all possible LLM operation modes.

dataflow.llms.api.get_providers()

Get all providers from the database.

Return type:

list[haive.dataflow.llms.api.llms.models.Provider]

async dataflow.llms.api.read_model(model_id)

Get a specific model by ID.

Parameters:

model_id (str)

async dataflow.llms.api.read_models(provider=None, capability=None)

Get all models with optional filtering.

Parameters:

• provider (str | None) – Filter by provider name

• capability (str | None) – Filter by capability (e.g., ‘vision’, ‘function_calling’)

async dataflow.llms.api.read_providers()

Get all LLM providers.

async dataflow.llms.api.recommended_models(task=None, vision=False, function_calling=False, audio=False, web_search=False)

Get recommended models based on capabilities and task requirements.

Parameters:

• task (str | None) – The type of task (chat, completion, embedding, etc.)

• vision (bool | None) – Whether vision capabilities are required

• function_calling (bool | None) – Whether function calling is required

• audio (bool | None) – Whether audio processing is required

• web_search (bool | None) – Whether web search is required