EmpirioLabs AI MCP Server

What is EmpirioLabs AI MCP Server?

The EmpirioLabs AI MCP Server is EmpirioLabs' official hosted MCP endpoint that lets any MCP client use the whole EmpirioLabs platform as tools: text models, media generation, web research, jobs, batches, GPU Cloud and hosted agents.

EmpirioLabs is an AI inference and integration provider. It hosts open-source models on its own GPUs, runs endpoints for proprietary models, and offers GPU Cloud instances and hosted agents, billed with pay-as-you-go credits or an optional plan with weekly allowances. The EmpirioLabs MCP server is remote, so there is nothing to install: you connect to https://mcp.empiriolabs.ai/mcp over Streamable HTTP and sign in.

How to use EmpirioLabs AI MCP Server?

You need an EmpirioLabs account with credits. Claude, ChatGPT and other OAuth clients open an EmpirioLabs sign-in page in the browser, where you review the request and choose Allow access. Editors and scripts can instead send an EmpirioLabs API key as a bearer token, which gives every tool without a sign-in step.

Claude Code

claude mcp add --transport http empiriolabs https://mcp.empiriolabs.ai/mcp

Run /mcp inside Claude Code to complete the sign-in the first time. To use an API key instead:

claude mcp add --transport http empiriolabs https://mcp.empiriolabs.ai/mcp --header "Authorization: Bearer ${EMPIRIOLABS_API_KEY}"

Claude Desktop and claude.ai

Open Settings → Connectors, choose Add custom connector, and enter https://mcp.empiriolabs.ai/mcp. Leave the OAuth client fields empty, click Connect, sign in to EmpirioLabs and choose Allow access.

Cursor

Add the server to ~/.cursor/mcp.json. Sign in through OAuth when Cursor prompts, or add an Authorization header with your key:

{
  "mcpServers": {
    "empiriolabs": {
      "url": "https://mcp.empiriolabs.ai/mcp"
    }
  }
}

VS Code

VS Code reads the same server from .vscode/mcp.json with a servers key:

{
  "servers": {
    "empiriolabs": {
      "type": "http",
      "url": "https://mcp.empiriolabs.ai/mcp"
    }
  }
}

The docs also list the server in the GitHub MCP Registry for VS Code and GitHub Copilot, and give setup for ChatGPT, Codex CLI, Grok, Perplexity, Le Chat, Gemini, Google Antigravity and the Vercel AI SDK.

Key features of EmpirioLabs AI MCP Server

• One endpoint for the whole platform: chat with every text model, generate images, video, audio, speech, 3D and transcripts, and run web search and research.
• Toolsets that double as OAuth scopes: models, text, search, media, jobs, agents, account, batch, gpu and hosted_agents, so the consent screen shows exactly what you grant.
• Profiles: add a profile such as /mcp/inference, or a +-separated list of toolsets, to the path to expose a smaller tool list, for example a connection that cannot generate media.
• Long-running jobs: media and agent tools return finished output URLs when work completes within the wait window (default 40 seconds, at most 50); otherwise they return a job_id for get_job.
• Same prices and limits as the API: tools use the published model prices, account limits and error messages.
• ChatGPT compatible: the server also provides the search and fetch tools ChatGPT uses for connectors and deep research.
• Revocable access: disconnect any connected app under Connected applications in the dashboard settings.

Use cases of EmpirioLabs AI MCP Server

• Pick a model by price: "List the cheapest video models on EmpirioLabs, then generate a 5 second clip of waves at sunset with the best value one."
• Transcribe and summarize: "Use EmpirioLabs to transcribe this recording, then summarize the action items."
• Track spending: "Check my EmpirioLabs balance and show what I spent on image generation this week."
• Provision GPUs: "Deploy an H100 on EmpirioLabs GPU Cloud with the vLLM template and tell me when it is ready."
• Cited research: "Run a web research task about the latest MCP specification changes and cite sources."

Troubleshooting EmpirioLabs AI MCP Server

• A generation returned a job_id instead of output: the work did not finish in the wait window. Call get_job (which can also wait) rather than starting the generation again.
• Output links stopped working: output URLs are signed and expire, so download anything you want to keep.
• A tool result says the account needs credits: tools draw on your EmpirioLabs credit balance; top up in the dashboard and retry.
• Tools do not appear in Google Antigravity: Antigravity uses serverUrl instead of url and does not report a failed connection. Check the key first, then ask the agent to list its MCP servers and tool counts.
• Too many tools for your client: connect to a profile path or a list of toolsets instead of the full /mcp endpoint.