Skip to content

Summary

Official WaveSpeed MCP server: search the live model catalog, read each model's real input schema, quote the price and run image, video, audio and 3D generation from any MCP client.

Features

  • Live model catalog search across image, video, audio and 3D
  • get_model_schema returns each model's real inputs before a run
  • run_model plus generate/edit image and generate/edit/extend video shortcuts
  • Free price quotes with get_price; balance, history, billing and usage tools
  • Hosted endpoint with OAuth-style approval, or local npx server with an API key
  • @path inputs upload local files automatically (local server)

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

  1. Open your agent's MCP configuration file.
  2. Merge the snippet below into it, filling in the environment variables with your own values.
  3. Restart the agent — the "WaveSpeed MCP" tools will be available.
{
  "mcpServers": {
    "wavespeed-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@wavespeed/mcp@latest"
      ],
      "env": {
        "WAVESPEED_API_KEY": "wsk_..."
      }
    }
  }
}

Hosted endpoint

This server is also hosted. A client that supports remote MCP servers can connect to it directly.

{
  "mcpServers": {
    "wavespeed-mcp": {
      "type": "http",
      "url": "https://api.wavespeed.ai/mcp"
    }
  }
}

Description

WaveSpeed's MCP server puts the WaveSpeed generative-media platform inside ChatGPT, Claude, Claude Code, Codex, Cursor, Cline and other MCP clients. Nothing is hardcoded to a fixed list of models: the catalog comes from WaveSpeed's live API, so a model the platform adds is usable the day it ships.

The design follows WaveSpeed's own CLI. list_models searches the catalog by text or modality, get_model_schema returns the exact inputs a model accepts so the agent reads before it writes, and run_model submits a run, rejects unknown input keys and returns output URLs. Five shortcut tools (generate_image, edit_image, generate_video, edit_video, extend_video) use the same models as WaveSpeed's chat page and fit the request to the chosen model. get_price quotes a run without charging, and account tools cover balance, run history, itemised billing, usage and file upload. Long generations return a prediction id at once and can be polled with get_prediction, so a dropped connection loses nothing.

There are two ways to run it. The hosted server at https://api.wavespeed.ai/mcp needs no install and signs in through a browser approval that creates a dedicated API key; clients that send headers can use Authorization: Bearer <key> instead. The local server is the @wavespeed/mcp npm package, run with npx and a WAVESPEED_API_KEY, and it can also upload local files referenced as @path inputs. Runs are billed to your WaveSpeed account at platform prices. The server code is MIT-licensed.

MCP: Wiro MCP

by Wiro AI

New

Official Wiro AI MCP server: search, price and run image, video, audio and LLM models, build and run multi-step workflows, and deploy and message AI agents from any MCP client.

Featured

Replicate's official MCP server: search thousands of hosted models, read their schemas, and run predictions on image, video, audio and language models from inside an agent.

MCP: MiniMax MCP

by MiniMax

New

MiniMax's official MCP server for its media APIs: text to speech, voice cloning and voice design, image generation and editing, and Hailuo video generation.

MCP: Ludo AI MCP

by Ludo.ai

Generate game-ready sprites, animations, 3D models, video, music and sound effects from your AI assistant through Ludo.ai's hosted MCP server.

Browse all MCP servers →