lurq

MCP server

Home

MarvinLabs vram_fit

Will a QLoRA fine-tune of gemma-3-4b-it fit your GPU? Verdicts from measured AMD and NVIDIA probes.

Endpoint
Answering
Works in
20 of 21 clients
Tools
3
Last checked
Oct 31, 2026

Does MarvinLabs vram_fit work in your MCP client?

Registry name org.marvinlabs/vram-fit, endpoint https://marvinlabs.org/mcp.

  • Claude Codeworks
  • Claude Desktopworks
  • Claude.aiworks
  • Claude API MCP connectorworks
  • ChatGPTworks
  • OpenAI Responses APIworks
  • Codexworks
  • Gemini CLIworks
  • Cursorworks
  • VS Code (GitHub Copilot)works
  • Windsurf (Devin Desktop)works
  • Zedworks
  • JetBrains AI Assistantworks
  • Junieunknownno source confirms Junie supports Streamable HTTP
  • Clineworks
  • Continueworks
  • Gooseworks
  • Lovableworks
  • Replit Agentworks
  • Bolt.newworks
  • v0works

How clients sign in

No credentials: it answered lurq's probe without any.

Tools

vram_fitvram_max_rankvram_probes

Get the setup for your client from your agent

This page is a daily snapshot. lurq's connect_check answers for the client you are wiring this server into, with the exact steps and the config to paste in that client's own format, and lurq mcp-pin tells you when the server changes. One command connects Claude Code, Cursor, VS Code and other agents:

npx lurqrun

Free to start. How connect_check works

From lurq's credential-free probe of this server's endpoint on Oct 31, 2026, and each client's own documentation. This page refreshes every 168 hours. How the probe works, and how maintainers opt out.