Connect an assistant

Five-minute demo

Connect OneInfer, sign in, then send these prompts in order. Open Console, Settings, and copy your Developer ID before the credits prompt. Chat and image generation spend credits.

1. List the OneInfer models I can use.
2. Send hello to a OneInfer chat model. Use the model_id from that list, not the display name.
3. How many OneInfer credits do I have? Use the developer id from Settings.
4. Generate an image of a red square with a OneInfer image model.
5. List my OneInfer GPU machines.

These three prompts should not call a tool: “Transfer $100 to another account.” “Delete my OneInfer account.” “Here is my API key. Paste it into the tool.”

Claude

Add a custom connector named OneInfer with the URL https://mcp.oneinfer.ai/mcp.

Open the Claude add-connector link, then sign in with your OneInfer account.

Claude Code

Run this in your terminal. --scope user adds OneInfer for all your projects.

claude mcp add --transport http --scope user oneinfer https://mcp.oneinfer.ai/mcp

Start Claude Code, type /mcp, choose oneinfer, then choose Authenticate. Sign in with your OneInfer account in the browser window that opens.

ChatGPT

1. Open Settings, then Plugins, and turn on Developer mode.
2. Choose Browse plugins, then the plus button, then Add custom MCP server.
3. Name it OneInfer. Set the URL to https://mcp.oneinfer.ai/mcp. Choose OAuth and dynamic client registration.
4. Create the connector, open a Work chat, type @, and select OneInfer.

Perplexity

Custom connectors need a paid Perplexity plan.

1. Open Settings, then Connectors, and add a custom connector.
2. Choose Remote. Name it OneInfer, set the URL to https://mcp.oneinfer.ai/mcp, and choose OAuth.
3. Sign in with your OneInfer account when Perplexity asks.
4. Turn on OneInfer in your thread’s sources, then send the demo prompts.

Codex

Run these in your terminal, then sign in with your OneInfer account in the browser window that opens.

codex mcp add oneinfer --url https://mcp.oneinfer.ai/mcp
codex mcp login oneinfer

Or add it to ~/.codex/config.toml yourself, then run the login command. The Codex IDE extension uses the same config.

[mcp_servers.oneinfer]
url = "https://mcp.oneinfer.ai/mcp"

Cursor

Add OneInfer to Cursor in one click, or add this to ~/.cursor/mcp.json for all projects, or .cursor/mcp.json for one project.

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

Then open Cursor Settings, then MCP, and sign in next to oneinfer.

VS Code (GitHub Copilot)

Add OneInfer to VS Code in one click, or add this server in .vscode/mcp.json. VS Code opens a browser so you can sign in. Then open Copilot Chat in agent mode.

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

Smithery

Install the published listing at smithery.ai/servers/oneinfer/mcp-server. Sign in with your OneInfer account when Smithery asks.

What you can ask for

The assistant can list models, run chat, generate an image or a video, read prices and credits, list or create GPU machines, use Studio storage, movies, music, and voiceover, and create a dedicated endpoint. Credits, GPU prices, GPU machines, and dedicated endpoints need the Developer ID from Console, Settings.

See all 24 tools by name on the OneInfer MCP page.

Chat, images, video, Studio movies, music, voiceover, GPU machine creation, and dedicated endpoints spend credits. The connector does not buy credits, create API keys, delete the account, or accept an API key as a tool argument.