LaunchDarkly's Developer plan is free with unlimited seats. Foundation, the first paid production tier, doesn't charge per seat at all — it meters two things: $10/month per service connection (beyond 5 included) and $8.33 per 1,000 client-side monthly active users. Enterprise and Guardian are both custom-quoted.
- LaunchDarkly removed per-seat charges from its current pricing model — a genuine structural change
- A lot of pricing content online still quotes the old $10-12/seat figure, which no longer applies
- Service connections count every running server-side SDK instance per environment — autoscaling can multiply this silently
- Client-side MAU bills on your highest-volume context kind, which can mean paying for devices, not just logged-in users
If you remember LaunchDarkly as a per-seat tool, that memory is now roughly two years out of date. The current model charges nothing for seats — Developer includes unlimited team members for free — and instead meters production usage on two dimensions that have nothing to do with headcount.
This distinction matters more than it might sound like at first. A team's LaunchDarkly bill today is driven by infrastructure architecture (how many server processes connect) and audience size (how many end users see a flag), not by how many engineers are on the account.
1. How much does LaunchDarkly cost?
Source: launchdarkly.com/pricing (verified) + third-party analysis, verified Sep 19, 2026.
Foundation is pay-as-you-go, billed yearly after a 14-day trial. There's no flat monthly number — your bill is $10/month per service connection beyond the 5 included, plus $8.33 per 1,000 client-side MAU beyond the 1,000 included, plus $5 per additional 1,000 AI runs beyond 5,000 included. Enterprise and Guardian add governance and release-automation features on top, entirely custom-quoted.
2. What each plan includes
| Feature | Developer | Foundation | Enterprise | Guardian |
|---|---|---|---|---|
| Unlimited feature flags | ✓ | ✓ | ✓ | ✓ |
| Unlimited seats | ✓ | ✓ | ✓ | ✓ |
| A/B testing & experimentation | ✓ | ✓ | ✓ | ✓ |
| SSO | ✕ | ✓ | ✓ | ✓ |
| SAML/SCIM, custom roles, approval workflows | ✕ | ✕ | ✓ | ✓ |
| Automated rollback on metric guardrails | ✕ | ✕ | ✕ | ✓ |
3. Cost calculator
4. Hidden costs to budget for
| Scenario | Annual cost | Change |
|---|---|---|
| Developer, unlimited seats | $0/yr | Base |
| Foundation, 10 service connections, 50K MAU | ~$5,600/yr | ($10×5 extra + $8.33×50) × 12 |
| Enterprise, 25 licenses (reported) | ~$38,000/yr negotiated | Plus a reported $30,000 platform fee |
Beyond the two core meters, one buyer report described a separate $30,000 Enterprise platform fee layered on top of negotiated per-license pricing — a fixed cost that doesn't show up when comparing per-connection or per-MAU rates alone. Client-side MAU also has a specific wrinkle: it bills based on your highest-volume context kind, which means a consumer app with 500,000 devices pays on that figure even if far fewer people are actually logged in and identifiable.
5. How LaunchDarkly pricing has changed
6. Which plan should you choose?
- For personal projects or evaluating LaunchDarkly before production use, Developer's free tier with unlimited seats and 1,000 MAU covers real testing.
- For production feature flag management at moderate scale, Foundation's pay-as-you-go model works well if you can forecast your service connection count and MAU.
- Teams needing approval workflows, custom roles, and SAML/SCIM should expect an Enterprise quote, and should specifically ask whether a platform fee applies on top of usage.
- Organizations wanting automated rollback tied to metric guardrails should evaluate Guardian directly with sales.
7. Is it worth the price?
For engineering teams that deploy frequently and need reliable, governed feature flag rollouts, LaunchDarkly's depth is a genuine differentiator over building flag infrastructure in-house. The usage-based model rewards teams with efficient, well-architected service connections and punishes over-provisioned or poorly-scaled infrastructure — which makes right-sizing your actual server-side connection count a real, ongoing cost lever rather than a one-time setup decision.
8. How to pay less
- Audit your actual service connection count before assuming your bill will match your architecture diagram — autoscaling and redundant replicas can inflate this meter without anyone deliberately deciding to.
- Understand which context kind drives your client-side MAU meter and whether it's counting devices, sessions, or identified users — this changes your real cost forecast significantly.
- Ask directly whether an Enterprise platform fee applies on top of your quote — this is a real, reported cost that isn't obvious from per-license negotiation alone.
- Re-verify any internal budget or vendor comparison built on the old per-seat pricing model — a genuinely common source of outdated planning given how recently this changed.
Per Vendr's tracked data across 196 purchases, the median LaunchDarkly contract runs $71,847/year — useful context, though actual cost varies enormously based on service connection count and MAU volume specific to your architecture.
9. Alternatives and what they cost
| Tool | Entry price | Cheaper when |
|---|---|---|
| GrowthBook | Free / custom | Predictable, warehouse-native pricing matters more than LaunchDarkly's broader observability suite |
| Split.io | Custom | A comparable feature-flag-focused alternative with different metering fits your architecture better |
| Unleash | Free (open source) / custom | Self-hosting is acceptable and predictable, controllable cost matters most |
10. The bigger picture: why LaunchDarkly bills drift
A vendor moving from seat-based to usage-based pricing is a structural shift that invalidates old budgeting assumptions overnight, and LaunchDarkly's transition is a clean example of exactly that. Zapro AI is built to flag this kind of pricing-model change across your vendor list before it becomes an outdated internal assumption.
11. The bottom line
LaunchDarkly's move away from per-seat pricing is a genuine structural change, and treating it as a simple price increase or decrease misses the point — the entire cost basis shifted to infrastructure and audience metrics. Model your actual service connections and client-side MAU realistically, and don't trust older pricing guides (including possibly ones from a year or two ago) that still describe the retired per-seat model.
FAQs
No. LaunchDarkly's current pricing model has no per-seat charge on any tier — Developer includes unlimited seats for free, and Foundation meters service connections and client-side MAU instead.
It's one server-side SDK instance connected to one environment. Multiple replicas of the same backend service, or the same service running in both staging and production, each count as separate billable connections.
It bills based on unique end users (or devices, depending on context kind) receiving feature flags via client-side SDKs each month — and specifically on whichever context kind has the highest volume, which can mean paying on device count even with fewer logged-in users.
No. Foundation is entirely usage-based: $10/month per service connection beyond the included 5, plus $8.33 per 1,000 client-side MAU beyond the included 1,000, with no flat subscription number.
The most common way is service connections multiplying through autoscaling or redundant deployments without anyone tracking the cost impact. A vendor spend platform that ties infrastructure changes to usage-based billing is the most reliable way to catch this.