DOCSEARCHbyAlgolia
Submit Docs

Eachlabs Documentation

website·Indexed 34 days ago

https://docs.eachlabs.ai/

A generative media platform providing access to 600+ AI models through a unified API. Features include the LLM Router with OpenAI-compatible endpoints for chat completions, text-to-speech, and transcription, as well as multi-step workflows, a Decisions API for structured answers, and webhook integration for prediction results.

Tokens
78.2K
Snippets
375
Records
525
Agent score
62%

What's inside Eachlabs

  1. Overview of each::sense AI agent tools

    The each::sense AI agent uses tools to perform actions. Tools are categorized into two types:

    • Non-terminal: The results are fed back to the agent for further processing.
    • Terminal: The execution completes the current task or sequence.

  2. Overview of each::storage managed file hosting

    each::storage is a managed file hosting service for media (images, videos, audio) used within the each::labs platform. It provides stable public URLs that can be passed directly into each::api predictions and each::workflows, eliminating the need for users to maintain their own storage buckets or CDNs. Files are stored in the United States and delivered via the cdn-us.eachlabs.ai CDN over HTTPS.

  3. Overview of each::labs generative media products

    each::labs provides a suite of tools for developers and AI agents to access 500+ production-ready AI models for image, video, audio, and 3D generation:

    • each::api: Direct model access via a single endpoint with async predictions and webhook delivery.
    • each::workflows: Multi-step pipelines for chaining models with parameter passing, versioning, and bulk execution.
    • each::sense: An AI agent interface (OpenAI-compatible) that handles natural language, auto-selects models, and generates media.
    • LLM Router: A single OpenAI-compatible endpoint providing access to 300+ LLMs from providers like OpenAI, Anthropic, and Google.
    • MCP Server: Integration for IDEs like Cursor, Claude Desktop, and Windsurf to generate media within the editor.

  4. Overview of LLM Router Audio API

    The LLM Router provides OpenAI-compatible endpoints for bounded speech-to-text and streamed text-to-speech. The API is currently in staged preview and enabled per environment; a 404 response indicates the capability is not yet enabled for your deployment. API keys must be provided via the Authorization: Bearer header; query-string keys are rejected for security.
    Base URL: https://api.eachlabs.ai/v1

  5. Overview of each::api

    each::api provides a single, unified interface to run predictions across 600+ AI models, including image generation, video, audio, and text. The general workflow involves browsing available models to determine required inputs, submitting a prediction request with specific model versions and parameters, and retrieving results via polling or webhooks.

  6. Overview of each::labs MCP Server

    The each::labs MCP server allows AI-powered editors and agents to access over 500 AI models for generating images, videos, audio, and 3D content. Once connected, users can perform tasks such as generating images (Flux, Gemini Imagen), videos (Veo 3, Kling 3.0, Sora 2), editing images (background removal, upscaling, face swap), generating audio (ElevenLabs TTS, Mureka), and running multi-step AI pipelines via each::sense.

  7. Overview of each::workflows

    each::workflows is an engine that allows developers to chain multiple AI models into automated, multi-step pipelines. It supports complex AI tasks where outputs from one step are wired as inputs to subsequent steps. Key capabilities include:

    • Parallel branching: Run independent tasks concurrently.
    • Conditional logic: Route execution based on step outputs using choice steps.
    • Parameter references: Use {{template}} syntax to wire step outputs to inputs.
    • Fallback configuration: Automatic retry with alternative models.
    • Bulk execution: Trigger up to 10 executions in parallel.
    • Webhooks: Receive notifications upon execution completion.
    • Versioning: Manage multiple versions of a workflow.
    • Public Sharing: Share workflows via direct links or public listings.

  8. Overview of the eachlabs Video API

    The Video API provides 44 programmatic capabilities for video processing, including transcoding, trimming, silence-aware splitting, captions/subtitle composition, watermarks, overlays, crossfades, audio replacement, ducking, mastering, and HLS/DASH packaging. It also includes four structured analysis capabilities (probe, scene_detect, silence_detect, audio_analysis) that return JSON instead of media files. The API uses the same API key and submit-and-poll flow as the eachlabs prediction API.

  9. Overview of each::sense generative media agent

    each::sense is an AI agent that acts as an intelligent layer for generative media. It provides a unified, OpenAI-compatible interface for generating images, videos, audio, and managing AI workflows. Instead of manually selecting models and parameters, users describe their requirements in natural language, and the agent automatically selects the best model from over 500 available options. Key capabilities include multi-modal support (text, image, video, audio, 3D), real-time SSE streaming with 18 event types, persistent session memory for multi-turn interactions, and natural language workflow creation.

  10. Understand Video API versioning and breaking change policy

    The Video API is versioned in its path (e.g., /v1/prediction). Changes that cannot be made compatibly result in a new version path rather than a silent edit to the existing one.

    Because the API sets additionalProperties: false for all capability parameter schemas, the engine rejects any request containing unknown keys before compute runs (at $0 cost). This means any change that narrows what the API accepts is considered a breaking change.

  11. Retrieve detailed information about a workflow

    Use the Get Workflow endpoint to retrieve comprehensive details about a specific workflow, including its metadata (name, status, tags) and a list of all its versions. You can identify the workflow using either its unique UUID or its URL-friendly slug.
    curl https://api.eachlabs.ai/v1/workflows/{workflowID} \
  -H "Authorization: Bearer YOUR_API_KEY"

  12. Retrieve detailed information about an AI model

    Use the Get Model endpoint to retrieve metadata, input schemas, and pricing estimates for a specific AI model using its unique slug identifier. This endpoint does not require authentication, although examples include an Authorization header for consistency with other API calls.
    curl "https://api.eachlabs.ai/v1/models/flux-1-1-pro"