{
  "id": 6620119,
  "title": "2B Gemma 4 Deployment with Cloud Run, NVIDIA L4, MCP SDK 2.x, and Claude Code",
  "url": "https://urgent.news/2026/09/10/2b-gemma-4-deployment-with-cloud-run-nvidia-l4-mcp-sdk-2-x-and-claude",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-10T18:31:29.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/gde/2b-gemma-4-deployment-with-cloud-run-nvidia-l4-mcp-sdk-2x-and-claude-code-4ml3"
  },
  "original_language": "en",
  "account": "This project is a DevOps/SRE assistant for deploying a Gemma 4 model using vLLM on Cloud Run with an NVIDIA L4 GPU. The deployment is managed through a single-file Python MCP server that simplifies tasks such as staging weights, deploying the service, checking health, benchmarking, and tearing down the deployment. The MCP server operates in a serverless environment where it scales to zero when idle. The MCP server relies on the MCP Python SDK version 2.x, which has undergone a recent migration. This migration necessitated updating the import statement and constructor from FastMCP to MCPServer within the server code, as well as installing the httpx2 package instead of the now-unavailable httpx. The project also introduces an MCP server that communicates with Claude Code, a tool used for model management.",
  "summary": "This article provides a step by step deployment guide for Gemma 4 E2B to a Cloud Run hosted GPU enabled system. A suite of Python MCP tools is built to simplify management of the vLLM hosted deployment with Claude Code. https://github.com/xbill9/gemma4-dev/tree/main/gpu-2B-cloudrun-devops-agent What is this project trying to Do? This project is a DevOps/SRE assistant for a Gemma 4 model served by…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}