LaunchDarkly Pricing in 2026: What You'll Actually Pay

I am building a feature flag product, so take this with the obvious grain of salt. The reason I am building one is that I kept watching small teams try to answer "what will LaunchDarkly actually cost?" and fail. Not because they cannot read a pricing page. Because the page meters two things that are not "using flags."

Numbers are list prices as of mid-2026. Verify against launchdarkly.com/pricing before you sign. If you want to skip the essay, run the calculator.

The two meters that matter

LaunchDarkly's paid Foundation bill is the intersection of:

1. Service connections. Roughly one SDK instance per environment: each backend service, worker, and client app, in each environment. About $10 per connection per month. Five services plus a web app across staging and production is 12 connections, $120/mo, before a single user shows up.

2. Client-side MAUs. Unique end users your client SDKs see, at $10 per 1,000 on monthly billing, $8.33 on annual. Grow from 25k to 100k users and this line goes 4×, even if you did not add a flag.

Seats are not the fight. Current LaunchDarkly tiers, including the free Developer plan, advertise unlimited seats. If a 2023 blog post told you the problem was $12/seat, that post is stale. I have a longer writeup on why the live hook is architecture.

The free Developer plan is generous on paper: unlimited seats and flags, many SDKs, an MCP server. It is sized as an on-ramp. Real traffic and a real service graph put you on the paid meters.

Worked example: 10 people, 25k MAU

• 5 backend services + 1 client app
• Staging + production = 2 environments → 12 connections × $10 = $120
• 25,000 client-side MAUs × $10 / 1k = $250

~$370/mo, about $4,400/yr, before experiments add-ons. The MAU line is most of the bill, and it only goes up.

Same shape on Betterflag: 25k MAU × 200 evaluations/MAU ≈ 5M evaluations = Launch at $24.99/mo.

Worked example: 100k MAU

MAUs at $8.33 to $10 per 1,000: $833 to $1,000/mo before connections. If you still have ~12 connections, add $120.

On Betterflag that is ~20M evaluations. Launch includes 5M, so the honest bill is Launch plus 15M overage at $84.99/mo, not $24.99. Scale at $99.99 includes 50M if you want the headroom and the audit log. Quoting Launch's included price against a 100k-MAU app is a lie, and it is the same sin as an unforecastable unit.

The conversion table lives on the calculator. Default is 200 evaluations per MAU per month. Cache more, pay less. Evaluate on every render, pay more. Change the slider.

Why it's built this way

None of this is an accident, and none of it is evil. LaunchDarkly sells to enterprises. MAU pricing maps to enterprise scale. Connection pricing maps to a streaming architecture that does cost them money. If you are a 500-person company with a platform team, this model is arguably fine.

The problem is that the meters tax the two things a startup does most: splitting services and getting users. A feature flag is a boolean lookup. Its cost to serve scales with evaluations, not with pod count or headcount. When the pricing meter and the cost-to-serve meter point in different directions, you pay for the gap.

What one meter looks like

Betterflag charges evaluations. Unlimited flags, seats, environments. Starter $9.99 (1M), Launch $24.99 (5M), Scale $99.99 (50M), overage published. No MAU math, no connection audit.

You may not pick Betterflag. Run this anyway:

1. Model the bill at 3× current traffic. If it scares you, that is the MAU meter.
2. Count connections honestly: (services + client apps) × environments, including preview.
3. Convert MAU to evaluations before you compare a $24.99 tier to a 100k-user app.
4. Ask where the audit log lives. If the answer is Enterprise, the kill switch has an upsell.

Full vs LaunchDarkly · feature flag pricing across vendors · calculator.