ModelPriceWatch.com
Last scan 2026-08-09 Models tracked 198 Providers 30 Cheapest paid Granite 4.0 Micro $0.017/Mtok in Every price links to its source

API plans

The full dataset is free and CORS-open — pricing, benchmarks, history, and rankings, no key required. Paid tiers exist for teams that need higher limits, a commercial license, bulk exports, or an SLA. Keys are issued by request (we'll set you up the same day).

Free

$0 /forever

For evaluation, prototyping, and low-traffic apps.

  • Yes100 requests / day
  • YesAll 8 endpoints
  • YesCurrent pricing data
  • YesBenchmark scores
  • YesNo API key needed
  • No commercial license
  • No SLA
  • No data exports
View docs

Startup

$99 /month

For teams building on LLM pricing data.

  • Yes100,000 requests / day
  • YesAPI key authentication
  • YesCommercial use license
  • YesBulk data exports (JSON, CSV)
  • Yes99.5% uptime SLA
  • YesPriority email support
Request a key

Enterprisecustom volume & requirements

$499 /month+

Custom volume & requirements

  • YesUnlimited requests
  • YesBulk data exports (JSON, CSV, Parquet)
  • YesHistorical data archive (all snapshots)
  • YesCustom benchmarks and scoring
  • Yes99.9% uptime SLA with credits
  • YesDedicated support + onboarding
Contact us

What each plan includes

side by side
Feature comparison of the Free, Pro, Startup and Enterprise API plans
Feature Free Pro Startup Enterprise
Daily requests 100 10,000 100,000 Unlimited
API key required No Yes Yes Yes
All 8 endpoints Yes Yes Yes Yes
Benchmark data Yes Yes Yes Yes + custom
Commercial use Yes Yes Yes
Data exports JSON, CSV JSON, CSV, Parquet
Uptime SLA 99.5% 99.9%
Support Email Priority email Dedicated

MCP server limits

per minute, not per day

The MCP server is metered per minute rather than per day. Agent traffic is bursty — a few tool calls while someone asks a question, then nothing — so a daily quota would either cut a real conversation short or do nothing about a client stuck in a retry loop.

MCP tool-call limits by plan
Plan Tool calls / minute Counted per Price
Free (no key required) 20 IP address $0
Pro 240 API key $19/mo
Startup 2,400 API key $99/mo
Enterprise Unmetered API key $499+/mo

Connecting, listing tools and pinging are counted separately and generously (120/min per IP, anonymous callers only), so a client reconnecting never spends its tool-call budget.

What happens when an MCP call is rate limited?

You get a normal MCP tool result flagged isError — not a transport error — carrying Retry-After, X-RateLimit-Limit and a plain-English explanation of the upgrade path. That is deliberate: an assistant can read a tool result and tell you what happened, whereas an HTTP-level failure usually surfaces as an opaque "server error". The body also links the free REST API, which serves the same data with no key, so a rate-limited agent still has somewhere to go.

Are these hard limits?

No, and we would rather say so than imply otherwise. The tool-call figures above are fair-use allowances to design against. Metering inside the Worker is deliberately loose and will usually let a burst through; sustained excess may be blocked at the network edge without warning — and that reservation covers GET /mcp, the discovery document, too. Because it changes only when we deploy, it ships an ETag and Cache-Control: max-age=300; fetch it once per session and reuse it rather than polling on a timer.

Do I need a key for the MCP server?

No. It stays open and unauthenticated at the free rate. A key only raises the ceiling — send it as X-API-Key or Authorization: Bearer to https://modelpricewatch.com/mcp. Request one here.

Frequently asked questions

Can I use the free tier in production?

Yes, but you must attribute modelpricewatch.com and link back to the site. The free tier is limited to 100 requests/day per IP and does not include a commercial license.

How do I get an API key?

Use our contact form and tell us your desired plan and use case. We'll issue your mpw_... key manually — usually same day. Include it in the X-API-Key header on all requests.

How are API keys managed?

Keys are stored in Cloudflare KV. Each key has a tier, email, creation date, and active status. Keys can be revoked at any time. Rate limits are tracked per-key per-day and reset at midnight UTC.

What happens when I hit my rate limit?

You get a 429 Too Many Requests response with a JSON body: {"error":"Rate limit exceeded","limit":100,"used":100,"tier":"free","reset":"midnight UTC"}. The response also includes Retry-After and X-RateLimit-* headers.

What counts as a request?

Each API call to any endpoint counts as one request. We recommend caching responses with a 4-hour TTL — data updates twice daily (09:20 & 21:20 UTC).

Can I get a refund?

Yes — 14-day money-back guarantee on all paid plans. No questions asked.

Do you offer custom integrations?

Enterprise customers get custom data delivery, bulk exports, and dedicated support. Contact us to discuss your needs.