Lunair's Pricing, Explained: Per-Second Billing on a Base44 App

Lunair sells video by the second. According to a Base44 case study, that model brought in $100K in its first two months and $8K in monthly recurring revenue (MRR, the subscription money that repeats every month). Guy Manzur built it on Base44, an AI app builder, and bills through Lemon Squeezy, a payments service.

Most AI SaaS apps charge a flat monthly fee. Lunair charges for the exact thing that costs it money: seconds of rendered video. This teardown looks at how the tiers work, where free ends, and what a first-time builder can borrow.

The three tiers

Here is the public pricing on lunair.ai, as listed in September 2026.

Do the math and a pattern appears. The Top-Up works out to $1.29 per second. The Monthly plan works out to $0.99 per second, because $29.70 divided by 30 seconds is $0.99. So subscribing is about 23% cheaper per second, and it adds the most-wanted perks.

Where free ends and paid begins

The free line is drawn around distribution, not creation. Free users get 20 seconds to try the whole product. But videos carry a watermark, stay public, and are shared through Lunair's own player.

That is a smart place to draw the line. A small business can test the output quality for free. The moment they want to post the video on their own site or ad account, they need a paid plan.

The Top-Up unlocks downloads. The Monthly plan is the only tier that lists "No watermark" and private videos. It also removes the per-project caps on edits and uploads. In other words, anyone using Lunair for real client or brand work is nudged toward the subscription.

Trial, refunds and the fine print

There is no time-limited trial. The free tier is the trial, measured in seconds instead of days. Because the seconds never expire, there is no countdown pressure.

The homepage does not show a refund policy or FAQ. We found a privacy policy link but no public terms or refund page. That is not unusual for a young product, but it is worth fixing. Clear refund terms reduce support email and chargebacks (when a customer asks their bank to reverse a payment).

What per-second pricing signals

Per-second pricing tells the buyer two things. First, you only pay for what you make. Second, the product is expensive to run, so the price tracks the cost.

AI video is costly to generate. Every second of output burns compute on rendering. A flat $29 plan with unlimited video could lose money on heavy users. Tying price to seconds protects margins.

Pure metered billing has a downside, though. Buyers hate surprise bills they cannot predict. Lunair avoids that by selling seconds in fixed bundles. You know the price before you click buy, and the balance simply runs down. It is closer to a prepaid phone plan than a utility meter.

The two revenue numbers also hint at buyer behavior. If only $8K a month is recurring, the rest of the first $100K likely came from one-time purchases. That is my reading, not a figure from the case study, which does not split the two. Base44 also reports 2,000 waitlist signups before launch and more than 5,000 signups since.

What to copy

• Price the unit that costs you money. If your app calls an AI model per image, per minute or per page, consider charging per unit.
• Make the subscription the best deal per unit. Lunair's monthly seconds are cheaper than one-time seconds, so regular users subscribe.
• Gate the output, not the trial. Let people see full quality for free. Charge when they want to download, remove branding or keep work private.
• Offer a no-commitment option. A one-time Top-Up catches people who hate subscriptions.
• Publish your refund terms. It costs nothing and builds trust.

If you vibe-code this

Usage-billed AI apps have a specific risk: credits are money. A few checks for this app type:

1. Only grant credits from a verified webhook. A webhook is the message your payment provider sends when someone pays. Check its signature before adding seconds. See Stripe webhook security; the same idea applies to Lemon Squeezy.
2. Deduct credits on the server, before rendering starts. Never trust a balance the browser sends. Use one database update that checks and subtracts in a single step.
3. Rate-limit the generate endpoint. Otherwise one script can queue hundreds of renders. See API rate limiting.
4. Enforce "private" on the server. Private videos need private storage and short-lived links. See storage bucket security.
5. Keep AI provider keys server-side. See environment variables.

Here is a signature check for a Lemon Squeezy webhook in a Next.js route:

import crypto from "crypto";

export async function POST(req: Request) {
  const raw = await req.text();
  const sig = req.headers.get("x-signature") ?? "";
  const digest = crypto
    .createHmac("sha256", process.env.LEMONSQUEEZY_WEBHOOK_SECRET!)
    .update(raw)
    .digest("hex");

  const a = Buffer.from(digest, "hex");
  const b = Buffer.from(sig, "hex");
  if (a.length !== b.length || !crypto.timingSafeEqual(a, b)) {
    return new Response("Invalid signature", { status: 401 });
  }

  const event = JSON.parse(raw);
  // Only now: add seconds to the user's balance
  return new Response("ok");
}

Building a usage-billed AI app and want a second pair of eyes on credits, webhooks and storage? Email me at [email protected] for a security audit.

Key takeaway

Lunair charges for the thing that costs it money, makes the subscription the cheapest way to buy it, and gates branding-free output rather than the trial. Any AI app with real per-use costs can copy that structure.

More case studies: vibe-coded apps making money. Before you launch: the vibe coding security guide.

Sources

• https://base44.com/blog/builder-spotlight-how-guy-manzur-built-an-ai-video-platform-and-hit-50k-in-30-days
• https://lunair.ai
• https://app.lunair.ai