About inferenceindexer-mcp
inferenceindexer-mcp exposes InferenceIndexer live and historical inference pricing to AI agents and MCP clients for model and provider comparisons. Alongside price lookup, it offers model recommendations and SIT index data.
Key features
- 01Recommend models under constraints
- 02Retrieve model prices, trends, and endpoints
- 03Search and inspect model details
- 04Compare model prices across providers
- 05Browse providers and pricing statistics
- 06Query SIT Composite index and history
Use cases
- 01Filter models by budget and quality needs
- 02Review a model's historical price changes
- 03Compare providers for the same model
- 04Track inference price index trends
MCP support
The MCP server gives clients tools for InferenceIndexer model pricing, historical data, provider comparisons, and the SIT index. It can run locally over stdio or connect remotely through Streamable HTTP.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://api.inferenceindexer.ai/mcpclaude mcp add --transport http desmartin01-inferenceindexer https://api.inferenceindexer.ai/mcp{"mcpServers":{"desmartin01-inferenceindexer":{"type":"http","url":"https://api.inferenceindexer.ai/mcp"}}}Open in Cursor{"servers":{"desmartin01-inferenceindexer":{"type":"http","url":"https://api.inferenceindexer.ai/mcp"}}}Open in VS Code{"mcpServers":{"desmartin01-inferenceindexer":{"type":"http","url":"https://api.inferenceindexer.ai/mcp"}}}{"mcpServers":{"desmartin01-inferenceindexer":{"serverUrl":"https://api.inferenceindexer.ai/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
10| Parameter | Type | Required | Description |
|---|---|---|---|
| zdr | boolean | No | |
| limit | integer | No | |
| modality | string | No | |
| reasoning | any | No | |
| context_min | any | No | |
| eu_sovereign | boolean | No | |
| budget_max_usd_per_m | any | No |
Details
From the project READMEInferenceIndexer MCP Server
Model Context Protocol server exposing InferenceIndexer's live + historical inference pricing as agent callable tools. Instead of an agent assembling inference pricing itself (slow, incomplete), it can call these tools to get complete pricing, historical trends, provider comparison, and the SIT index.
Tools
| Tool | Description |
|---|---|
recommend_models | Call this first: rank models under your constraints (budget, context, ZDR, quality floor) with ready-to-call endpoint configs |
explain_model | Everything about one model in one call: pricing, trend, endpoints, privacy flags |
search_models | Search/list models with current pricing (by text, tier, sort) |
get_model | Full detail + current price for one model |
get_model_history | Historical price trends for one model (the differentiator) |
list_providers | All providers with model counts + price stats |
get_provider | Detail for one provider (models, tiers, price range) |
get_composite_latest | Current SIT-Composite index value + tier breakdown |
get_composite_history | SIT-Composite index history / trend |
compare_providers | Price of one model across all providers that host it |
Config (env)
II_API_BASE- InferenceIndexer API base (defaulthttps://api.inferenceindexer.ai)II_API_KEY- Optional bearer key (public server works without one; a key raises rate limits)II_SSR_SECRET- Optional first-party secret for self-hosted/higher-tier use only. Not set on the public server.
Run
stdio (recommended for local/agent-run):
uv sync
uv run inferenceindexer-mcp
Serve over HTTP/SSE (for remote clients, e.g. the hosted endpoint at https://api.inferenceindexer.ai/mcp):
uv run inferenceindexer-mcp --transport streamable-http --port 8899
Connect an agent / MCP client
Claude Desktop / generic MCP client (stdio):
{
"mcpServers": {
"inferenceindexer": {
"command": "uvx",
"args": ["inferenceindexer-mcp", "--transport", "stdio"]
}
}
}
Updated Oct 2026. Hosted (no install): https://api.inferenceindexer.ai/mcp - add "url": "https://api.inferenceindexer.ai/mcp" to mcpServers for streamable-http clients.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.