Loomal

Free vs Paid Weather Services APIs for AI Agents the best free upstreams in any category.

Weather is blessed with publicly funded data: NOAA, national met offices, and Open-Meteo give agents remarkable coverage for nothing. Paid weather access earns its keep only at the edges — commercial licensing, guaranteed delivery, and specialized models.

No category has a stronger free tier than weather. Taxpayer-funded agencies publish forecasts, alerts, and model output openly: the NOAA-focused servers here expose hurricane tracks (NHC MCP Server), ocean forecasts (NOAA OFS), storm surge (STOFS), and wave models (WW3), while Weather MCP Server and Weather Data MCP Server build on Open-Meteo and NOAA for general forecasts, air quality, and radar.

So the honest framing for this category: free is the default, and paid has to justify itself. Sometimes it does.

How far free weather data goes

Very far. National met offices publish openly across much of the world — vedurstofa-mcp serves Icelandic forecasts, warnings, and earthquakes from vedur.is; mcp-server-taiwan-weather wraps Taiwan's Central Weather Administration; the NOAA family of servers covers US marine and storm modeling in depth. Open-Meteo aggregates global model data with a free tier for non-commercial use.

For an agent answering 'will it rain tomorrow in Reykjavik,' there is no reason to pay anyone. The data is public, the servers are open source, and coverage is genuinely global.

Where the free tier has fine print

Free weather upstreams carry conditions worth reading. Some, like Open-Meteo, distinguish non-commercial from commercial use — check the current terms before shipping a product on a free tier. Government APIs are reliable but unprioritized: no SLA, occasional maintenance windows, and rate limits designed for polite use, which a fleet of agents polling forecasts can exceed.

And free sources serve standard products. Hyperlocal nowcasting, route-specific forecasts, or derived indices for agriculture — the territory agrobr touches with Brazilian crop and climate data — often involve processing someone has to fund.

What paid weather access is actually for

Paid weather endpoints make sense in three cases: commercial licensing (your product resells forecasts and the free tier's terms exclude that), guaranteed delivery (your agent's decisions are time-critical and you need someone accountable for uptime), and value-added processing (custom models, blended sources, domain-specific outputs).

x402 fits these as per-call purchases: the agent requests a forecast, gets an HTTP 402 with the price, pays in USDC, and receives the data — settled on Base in about two seconds, from $0.01 per call. For a logistics agent checking marine conditions per voyage rather than per minute, per-call pricing is exactly proportional to need.

Choosing for this category

Default to free: the open servers over NOAA, Open-Meteo, and national met offices cover most agent use cases at zero cost. Pay when your use is commercial under the upstream's terms, when you need an operator on the hook for delivery, or when the forecast itself is a processed product rather than public data.

Loomal's Weather Services category lists all 59 live servers with descriptions and x402 pricing where maintainers have configured it, making the free-by-default landscape easy to scan.

Frequently asked questions

Should my agent use a free or paid weather MCP server?

Default to free — weather has publicly funded upstreams like NOAA and Open-Meteo, and open servers like Weather Data MCP Server expose them well. Pay only for commercial licensing needs, delivery guarantees, or processed products beyond the standard public forecast.

Is Open-Meteo or NOAA data really free for my product?

NOAA data is US public domain. Open-Meteo offers a free tier with conditions — historically oriented toward non-commercial use — so check their current terms before building a commercial product on it. The MCP servers wrapping these sources are typically open source either way.

How does pay-per-call compare to a subscription for weather data?

If your agent checks weather per decision — per voyage, per delivery route, per event — x402 per-call pricing in USDC from $0.01 matches cost to value exactly. Subscriptions suit continuous high-frequency polling, which most agent workloads aren't.

Are paid weather endpoints more accurate than free ones?

Not inherently — many paid products are built on the same public model output. What paid access can add is processing (hyperlocal, blended, domain-specific) and accountability for delivery. For raw forecasts, the free public sources are the reference.

Run a Weather Services MCP server?

Claim your listing, set a per-call USDC price, and let AI agents pay for every call over x402.

List it on Loomal