Developers

MCP: pricing in your AI client

Connect your AI client to the same pricing source and calculation engine as this website. No API key is required.

{
  "mcpServers": {
    "modelcostcomparison": {
      "url": "https://modelcostcomparison.com/api/v1/mcp"
    }
  }
}

Try: “Use ModelCostComparison to find reviewed model offers and compare 1 million input tokens and 200,000 output tokens, with no cache and at most 8,000 input tokens per request.”

• list_models: find exact model and tariff IDs.
• get_model_pricing: inspect rates, sources and review dates.
• calculate_cost: estimate input, output and cached-input token costs.
• compare_model_costs: compare selected offers for the same token usage.
• find_cheapest_models: rank eligible offers for your input, output and cached-token workload.
• list_providers: browse providers, model counts and reviewed pricing ranges.

Integration repository · Official MCP Registry entry

Measured by MCPmetrics

Ask your client to list offers first, then provide input and output token counts. Cached input is included in total input. For larger workloads, also provide the maximum input tokens in a single request. Expired or unreviewed offers cannot be calculated.

Results are text-token subtotals. Cache writes, storage, tools, search and taxes are excluded. Comparisons report unsupported offers separately; they do not rank model quality.

Quickstart

1. Read the agent index: /llms.txt
2. Fetch OpenAPI: /openapi.json
3. Call the versioned advisor:

   curl -sS -X POST https://modelcostcomparison.com/api/v1/models/advisor \
     -H 'Content-Type: application/json' \
     -d '{"message":"support chatbot 50k messages/month"}'

4. Or install the CLI: npx modelcostcomparison advisor "…"

API keys

Documented public surfaces (datasets, OpenAPI, /api/v1/models/advisor) do not require an API key. The models advisor is rate-limited per client. The pricing-refresh cron endpoint requires Authorization: Bearer <CRON_SECRET> and is not for public clients.

Sandbox

Production hosts the public sandbox: download pricing JSON and benchmarks JSON, and POST the rate-limited advisor. No separate staging credentials are required for read-only integration tests. API index: GET /api/v1.

Versioning and deprecation

Stable REST operations use major-version URL paths beginning with /api/v1. Breaking changes ship under a new major path (/api/v2). Additive, non-breaking changes may appear on the current major without a path bump.

Deprecated operations return Deprecation: true (RFC 9745) and a Link rel="successor-version" header pointing at the replacement. When a removal date is set, responses also include Sunset (RFC 8594).

Canonical advisor path: /api/v1/models/advisor. Legacy alias /api/mcc-model-advisor remains available and returns Deprecation: true plus Link rel="successor-version".

Errors

REST API failures return RFC 9457 application/problem+json with stable code, detail, optional resolution, and a nested error object for older clients. Wrong methods return 405 with an Allow header — never an empty body. MCP uses JSON-RPC and structured tool-result errors.

Official CLI

Package: modelcostcomparison on npm (homepage https://modelcostcomparison.com).

npm i -g modelcostcomparison
modelcostcomparison --help
modelcostcomparison pricing
modelcostcomparison advisor "RAG chatbot with high cache"
modelcostcomparison openapi
# optional alias: alias mcc=modelcostcomparison

Machine-readable links

• OpenAPI
• RFC 9727 API catalog
• /developers/llms.txt
• /llms.txt