Use Open-Meteo until you can't

I run a paid weather API and I am about to spend a blog post recommending the free one. Open-Meteo is the right starting point for a soaring side project, full stop. If you are an indie dev or a pilot who codes and you want to build a flight-planning tool over the weekend, install their client library, pull CAPE and Lifted Index and PBL height from the GFS or ICON-EU endpoint, and ship. You do not need us yet. You might never need us. That is fine.

Open-Meteo is open source, free for non-commercial use up to 10,000 calls a day, well-documented, and it already serves a bunch of pilot tools - Meteo-Fly and FlyDecision are the two I see cited most often. It exposes the stability indices a thermal app actually wants (CAPE, Lifted Index, Convective Inhibition, PBL height) by re-serving them from upstream NWP models. The DX is excellent. You will be querying it inside ten minutes.

There is no 'gotcha' coming in this post. It really is that good for the use case. What I want to flag is where the limits actually are, so you can plan around them rather than being surprised by them six months in.

The first limit is the fields themselves. Open-Meteo passes through what the upstream NWP gives it. So you get CAPE, you get Lifted Index, you get PBL height - generic stability indices. You do not get wstar as an hourly variable (they expose 'maximum updraft' only as a daily aggregate). You do not get a glider-capped soaring height. You do not get a composite day rating. You do not get a horizontal convergence field. If your app needs any of those, you will end up deriving them yourself from the raw fields, which is doable but it is now your code doing physics that someone else could be doing for you.

The second limit is the licensing. Free tier is non-commercial only. The moment you put ads on, take a subscription, or sell the app to a club, you need a commercial Open-Meteo plan - and at that point you are paying for a general weather API to do a soaring-specific job. That is the point where it stops being a question of 'free vs paid' and starts being a question of 'which paid tool is the right fit'.

The third limit is the model itself. Open-Meteo's free tier serves global NWP at 13 km (GFS) or regional at 2 km (ICON-D2 over Germany and the Alps, HRRR over North America). The European regional models are excellent. But there is nothing in there that was tuned for daytime convective conditions in the mid-latitudes specifically. The PBL height you get is the upstream model's PBL height, with whatever bias that scheme has. The convective fields are the upstream model's convective fields. For a hobby project, this is more than enough. For something where pilots will route XC flights off your output, you might want a model whose default failure modes are biased toward soaring-safe directions rather than generic-weather-safe directions.

The right test is honest. If your project is non-commercial, can live with generic stability fields, and you are happy deriving anything fancier, Open-Meteo is the answer and we recommend it. If your project is commercial, needs native RASP-style fields (wstar, BL top, soaring ceiling, convergence) out of a soaring-tuned model, and wants a free tier you can ship in production, the head-to-head comparison page lays out where Convek fits instead.

The point of saying any of this out loud is that I think the soaring-API market is much better served by being honest about who fits where than by pretending everyone is a competitor. Open-Meteo is genuinely great. Use it until you can't.