Provides MCP tools to inspect FreeModel key/session health, discover available models, get model recommendations, and run tasks using stored provider API keys.
freemodel_key_health — check subscription status, platform health, and recommended model.
freemodel_status — get a session summary: active model, platform, subscription, healthy count.
freemodel_models — list all AI models available to you across platforms.
freemodel_recommend — recommend 2–3 models for a given task with reasons.
freemodel_run — execute a task using a specific platform and model, with optional system prompt, max_tokens, temperature, and reasoning_effort.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@FreeModel MCP help me write a Python script to scrape a website"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An MCP server for the FreeModel gateway: look up the model catalogue, see what your key is allowed to call, and send a prompt — from inside Claude Code or any other MCP client.
npx freemodel-mcpTool | Calls | Key needed |
|
| no — the catalogue is public |
|
| yes |
|
| yes |
|
| yes |
freemodel_models — the catalogue, the same endpoint the counts on the site are derived from. Filter by modality (chat, image, video, tts, asr, embedding, rerank) or by id substring.
freemodel_tiers — the three tier aliases and what is in each band: fm-v1-lite (free, any account), fm-v1-standard, fm-v1-pro. These are pools rather than single models: asking for a tier means "any model in this band", which is what lets a request survive one provider going down.
freemodel_account — subscriptions and when they expire, which tiers this key may call, usage so far.
freemodel_chat — one prompt, one answer. Defaults to fm-v1-lite; pass any id from freemodel_models to pin a specific model.
Related MCP server: ToolRoute
Get a key at freemodel.online/console → API keys.
{
"mcpServers": {
"freemodel": {
"command": "npx",
"args": ["-y", "freemodel-mcp"],
"env": { "FREEMODEL_KEY": "sk-your-key" }
}
}
}OPENAI_BASE_URL = https://freemodel.online/v1
OPENAI_API_KEY = sk-your-keygit clone https://github.com/yummy342/freemodel-mcp.git
cd freemodel-mcp && npm install{
"mcpServers": {
"freemodel": {
"command": "node",
"args": ["/path/to/freemodel-mcp/server.js"],
"env": { "FREEMODEL_KEY": "sk-your-key" }
}
}
}Variable | Default | Meaning |
| — | Your gateway key. Only the keyed tools need it. |
|
| API root. Change it to point at another deployment. |
The free tier needs no card. fm-v1-standard and fm-v1-pro need a subscription, which is bought on the account portal at alli.website — without it those calls are refused with a 402, never billed silently.
No routing across your own provider keys. Version 1.x called the /api/gateway/agent/* endpoints, which pick among the provider keys attached to an account and answer No platforms configured when none are attached. This package talks to the gateway's own catalogue and tiers instead, so a plain key is enough.
No streaming yet. freemodel_chat waits for the full answer.
No embeddings or image calls. Those exist on the gateway (auto/embed, auto/image, and the rest), but this package does not wrap them.
MIT
Check API key health: subscription status, platform health, and recommended model. Use FIRST before task routing.
Inputs: no named parameters.
List all AI models available to you across platforms. Returns platform and model names.
Inputs: no named parameters.
Recommend the best AI model for a given task. Analyzes your task and recommends 2-3 models with reasons.
Inputs: task (required).
Execute a task using a specific AI model from a specific platform. Uses your stored API keys.
Inputs: task (required), model (required), platform (required), system, +3 more.
Get FreeModel session summary: active model, platform, subscription, healthy count.
Inputs: no named parameters.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 5 tools
Each tool has a clear primary purpose: health checks, session status, model listing, recommendation, and execution. However, freemodel_key_health and freemodel_status both touch on health/status, which could cause slight confusion, though their descriptions distinguish them.
All tool names follow the consistent pattern freemodel_<verb_noun> (e.g., freemodel_key_health, freemodel_models). The verb is implicit in some (models, status), but the prefix unifies them perfectly.
Five tools is well-scoped for a model routing/execution server. Each tool covers a distinct step in the workflow, and nothing feels redundant or missing.
The tool surface covers the full lifecycle from health check and status to listing models, getting recommendations, and executing tasks. There are no obvious dead ends; the workflow is clear and self-contained.
AI model routing on your own vendor keys: pick the best model per prompt, or route and run it.
Real-time, drift-verified LLM pricing across 31 providers with cost+quality model routing.
Sourced AI-model pricing and capability data — compare and route to the cheapest capable model.
AI routing, memory, guardrails, and governance. Routes across Claude, GPT, Gemini.
