> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.empiriolabs.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.empiriolabs.ai/_mcp/server.

# Concepts

The EmpirioLabs API is organized around a few core concepts that appear throughout the docs and API reference.

## Core concepts

#### Models

211+ models across text, image, audio, video, and document modalities. Each model's per-model page lists its accepted input/output modalities, parameters, pricing, and supported endpoints.

#### Credits

The prepaid balance on your account. API usage debits credits based on the model's pricing: per token, per request, per image, per audio-second, or per video-second depending on the model type.

#### API Keys

Bearer tokens that authenticate your requests. Keys use the `sk-empiriolabs-` prefix. Each account can hold up to 50 keys (adjustable by contacting support).

#### Limits

Per-account throughput, API-key, and saved-playground-chat caps. Defaults are 50 RPM, 2,000,000 TPM, 50 API keys, and 50 saved chats.

## How it works

1. **Sign up and add credits.** Create an account, add funds from the Billing page, and generate an API key.
2. **Browse the model catalog.** Call `GET /v1/models` to see every available model with its pricing, context window, and supported parameters.
3. **Send requests using familiar formats.** Use `/v1/chat/completions` (OpenAI-compatible), `/v1/responses` (OpenAI Responses-compatible), or `/v1/messages` (Anthropic-compatible) to interact with any model.
4. **Monitor usage and spend.** Each request is logged with token counts, cost, and latency. Check the dashboard for real-time usage data.

## Compatibility layer

EmpirioLabs does not require you to learn a new API format. The platform exposes endpoints that match the OpenAI and Anthropic request and response shapes, so you can:

* Drop in the OpenAI Python/Node SDK by changing `base_url` and `api_key`
* Use the Anthropic SDK by pointing it at the EmpirioLabs `/v1/messages` endpoint
* Switch between models from different providers without changing your integration code

See [Compatibility](/compatibility) for endpoint details and examples.

## Key design principles

* **Unified access**: one API key, one base URL, 211+ models from multiple providers
* **Prepaid credits**: no surprise invoices; top up when you need more capacity
* **Provider-compatible formats**: existing OpenAI and Anthropic code works with minimal changes
* **Transparent pricing**: every model lists its pricing units and rates in the catalog API