groq

The Groq MCP Server exposes Groq model capabilities to MCP clients, enabling fast inference for vision, speech, and general prompts through Groq's models. It integrates with the MCP framework so clients can request tasks such as image understanding, speech synthesis and transcription, text-to-speech, and batch processing of large prompt sets. You can leverage supported agentic tools, vision analysis, and batch features to build interactive apps or automate workflows via MCP clients like Claude Desktop or other Groq-compatible clients. To start using it, configure your MCP client to point to the Groq MCP server configuration, ensure you provide your Groq API key, and then issue prompts or tool calls that map to Groq capabilities (e.g., TTS, STT, vision analyses, or code generation tasks) through the MCP interface.

Prerequisites:

• Python environment access and uv installed (as described by Groq's setup instructions).
• A valid Groq API key.

Option A: Quick setup with uv (recommended)

1. Install uv if you haven't already:
   • curl -LsSf https://astral.sh/uv/install.sh | sh
2. Clone the repository and enter it:
   • git clone https://github.com/groq/groq-mcp-server cd groq-mcp
3. Install dependencies and create the MCP configuration via uv:
   • uv venv
   • source .venv/bin/activate
   • uv pip install -e ".[dev]"

Option B: Manual installation

1. Clone the repository and create a virtual environment with uv.
2. Activate the environment and install dependencies:
   • uv venv
   • source .venv/bin/activate
   • uv pip install -e ".[dev]"
3. Copy and configure .env with your Groq API key and any needed paths.

Configure your MCP client to use the Groq MCP server as shown in the example configuration (mcp_config) and start the MCP client to connect to this server.

Tips and common notes:

• Ensure GROQ_API_KEY is kept secret and never exposed in public configs.
• BASE_OUTPUT_PATH is optional; if omitted, defaults may apply depending on the client (often a desktop output directory).
• If using Claude Desktop, enable Developer Mode as described in the Groq README to access MCP functionality.
• The Groq MCP server integrates with various MCP clients; for other clients, install groq-mcp via uvx or pip and generate the appropriate config.
• For batch tasks, you can leverage the groq_batch.sh and related scripts in the repository to process large prompt sets efficiently.
• When troubleshooting, consult the logs in your platform’s standard Claude/Groq log locations and ensure your API key has the required permissions.