The free key proves your parser against real REST responses, but it cannot show you the two things that decide most upgrades: the SSE drop stream and the raw WebSocket. For those, the provider offers a three-day full demo on request, and three days is enough if you arrive with a written checklist. Here is the checklist.
What the demo includes
The demo opens SSE, the raw WebSocket and a REST allowance of 10 requests a second, as the free key page and the pricing page both state, checked 2026-09-29. It exists to answer one question: which transport does your application actually need? Treat it as a measurement exercise, not a preview.
Ask for it once your REST code already works on the free key. The demo days are too short to spend the first one fixing a parser.
Day one: prove REST parity
Run your existing free-key code against the demo key. It should work unchanged, because the endpoints and the response envelope are the same across plans. What changes is the rate: at 10 requests a second you can finally measure your real polling shape. Log the requests per second your loop uses at peak, because that number picks the plan later.
Also watch how your client behaves near the ceiling. Push toward the limit, confirm your backoff honors the Retry-After header, and check that your logs would distinguish a quiet board from a stalled pipeline.
Day two: hold the SSE stream open
Open the drop stream and leave it running for a full day. You are measuring three things: the alert volume at your intended min_drop, the share of alerts your rules would actually act on, and how your reader handles the boring parts, meaning quiet periods and reconnects.
Send a few alerts all the way through your relay or model. A demo that ends without one alert reaching its destination has not tested anything.
Day three: decide about raw frames
Open the raw WebSocket only if you suspect you need it. The first frame on each stream is a snapshot and everything after it is an update, so ask the real question: do you have the reconstruction, recovery and monitoring capacity to build your own state from frames? The WebSocket guide lists what that work involves. If the answer is no, that is a good answer, and it saves the add-on fee every month.
Turn the days into a plan
The plan guide maps four loops to four plans, and your notes now say which loop is yours. If the stream sat unused, you are a polling application: the REST snapshots plan at 10 requests a second, $99 a month, is the fit. If the stream was the product and polling was incidental, the SSE plan at $99 a month fits. If both ran hard, the combined plan at $149 a month exists for you. If 10 a second throttled you, the high-volume plan at 30 a second is $229 a month. All figures as listed on the pricing page, checked 2026-09-29.
Whatever you pick, take one month first. Term discounts of up to 16% at three months and up to 24% at six reward a loop that has already proven itself, not one you hope to write. The access checklist carries you from decision to first verified request.