MCP server that lets Claude Code delegate tasks to GPT-5.6 (Codex CLI), DeepSeek, GLM, OpenRouter and local LM Studio models — plus a 198-run benchmark of all of them
Delegate tasks from Claude Code to other companies' models — without leaving the app.
This is a small MCP (Model Context Protocol) server that acts as a "waiter" between your main coding agent and every other model you have access to. Claude Code stays the orchestrator; the waiter takes an order to whichever kitchen you point at:
- Codex CLI → GPT-5.6 Sol / Terra / Luna via your ChatGPT subscription (no API cost) - DeepSeek (DS4 Flash / Pro) via direct API - z.ai (GLM 5.2) via the coding-plan subscription - OpenRouter → anything in their catalog - LM Studio → local models on your machine or another box on your LAN, for free
The same pattern works for any MCP-capable agent — nothing here is Claude-specific except where it's registered.
Tool What it does ------ listmodels Returns the menu: every enabled model with its exact id and provider status (missing key, offline local server, etc.) delegatetask Sends a self-contained task to the chosen model and returns its answer, tagged with origin and token usage
Then ask Claude: "use the listmodels tool and show me the menu".
- config/models.json — which providers exist, their base URLs, and which models are enabled. Adding an OpenAI-compatible provider is one JSON block; enabling a model is one line in its models array. - .env — API keys only. Never in models.json, never in code. The server reads models.json fresh on every call (edit and it applies immediately); .env is read at startup (restart the server after adding a key). - Local control panel — npm run panel opens a localhost page (http://127.0.0.1:4747) to manage keys a
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/dpmadsen--multimodels-mcp.html)
ChatGPT thinks. Codex works. Use ChatGPT as the planning brain while keeping the Codex harness.
shinthink/blitzstrike⚡ Blitz Strike — a universal MCP penetration-testing toolbelt. Structured methodology: reconnaissance & attack-surface mapping, source-to-sink analysis, and live validation. 57 escalation chains, 130-
0xwilliamortiz/openclaude-improvedruns anywhere. uses anything
halofyai/halofyHalofy is the open access and governance layer for AI agents across your organization. Identity, policy, provenance, audit, and signed erasure.
CaptureGrubEnchant/SolidWorksSolidWorks MCP Server connects an AI assistant to a running SolidWorks instance. Sketch, extrude, fillet, export STEP/STL, generate macros.
arielshad/3d-asset-server3d assets api and mcp
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.