What is the DemandSphere MCP server?

It is a public Model Context Protocol server that exposes DemandSphere Radar research data to AI agents. Agents can query frontier AI model benchmarks, pricing, and capabilities, calculate token costs, and search Google algorithm update history.

Do I need authentication to use the MCP server?

No. The server is open and requires no API keys or authentication. Data is licensed under CC BY-NC 4.0.

What tools are available?

Five tools: get_ai_models (list/filter models), search_ai_models (text search), calculate_model_cost (pricing calculator), get_algorithm_updates (list/filter Google updates), and search_algorithm_updates (text search updates).

How do I connect my AI agent to this MCP server?

Add the server URL to your MCP client config. For Claude Desktop or Claude Code, add it to your MCP servers configuration with the URL https://www.demandsphere.com/mcp. No API key needed.

MCP Server

Endpoint

        https://www.demandsphere.com/mcp
        
      

MCP JSON-RPC 2.0 over HTTP. No authentication. No rate limiting. Powered by Cloudflare Workers.

License

CC BY-NC 4.0 View license

Attribution required - credit DemandSphere
Non-commercial use only
Commercial use requires written permission

Quick Start

Claude Desktop / Claude Code

{

  "mcpServers": {

    "demandsphere-radar": {

      "url": "https://www.demandsphere.com/mcp"

    }

  }

}

curl

curl -X POST https://www.demandsphere.com/mcp \

  -H "Content-Type: application/json" \

  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

Available Tools

get_ai_models AI Models

List and filter frontier AI models. Returns name, provider, release date, parameters, benchmarks, and pricing. Sorted by release date descending.

          Parameters: provider? type? access? after_date? before_date? limit?
        

get_ai_model_by_id AI Models

Get a single model's full record by ID. Includes all benchmarks, pricing, download links, and source URL.

          Parameters: id (required)
        

search_ai_models AI Models

Search models by name, provider, or description.

          Parameters: query (required) limit?
        

calculate_model_cost AI Models

Calculate token costs. Defaults to 1M input + 1M output if omitted (baseline comparison).

          Parameters: model (required) input_tokens? output_tokens?
        

get_providers AI Models

List all tracked providers with model counts and latest release dates. Use to discover valid provider names for filtering.

          Parameters: none
        

get_algorithm_updates Algorithm Tracker

List and filter Google algorithm and AI search updates from 2000 to present.

          Parameters: type? year? after_date? before_date? limit?
        

search_algorithm_updates Algorithm Tracker

Search algorithm updates by name or description.

          Parameters: query (required) limit?
        

get_server_info Meta

Server metadata: data freshness timestamps, model/provider/update counts, version. Use to verify recency.

          Parameters: none
        

Example: Search for "opus"

Request

{

  "jsonrpc": "2.0",

  "id": 1,

  "method": "tools/call",

  "params": {

    "name": "search_ai_models",

    "arguments": { "query": "opus" }

  }

}

Response (abbreviated)

{

  "query": "opus",

  "count": 3,

  "models": [

    {

      "name": "Claude Opus 4.7",

      "provider": "Anthropic",

      "released": "2026-04-16",

      "pricing": {

        "input_per_mtok": "$5",

        "output_per_mtok": "$25"

      }

    }, ...

  ]

}

Discovery

AI agents can discover this MCP server through standard discovery mechanisms:

        <link rel="mcp">
        In every page <head>
      
        /.well-known/mcp.json
        Server metadata and tool list
      
        /.well-known/mcp/server-card.json
        Extended server card with data source descriptions
      
        /llms.txt
        MCP server documented in site overview

Data Sources

AI Frontier Model Tracker

57 models across 15 providers. Benchmarks, pricing, architecture, and capabilities.

View tracker →

Algorithm Update Tracker

173 Google algorithm and AI search updates from 2000 to present.

View tracker →

MCP Server

Track how these models see your brand