Originally published at devtoolpicks.com
LaunchDarkly is the name everyone reaches for when they hear "feature flags." It's also the reason a lot of indie hackers go looking for something else.
The free Developer plan sounds generous until you read the caps: 1,000 client-side monthly active users and 5 service connections. Cross either line and you land on the Foundation tier, which is usage-based at roughly $10 to $12 per service connection per month plus about $8 to $10 per 1,000 client-side MAU, with experimentation and observability billed on top. The median LaunchDarkly buyer pays around $72,000 a year. That's a reasonable number for a funded team with a platform org. It's absurd for a solo founder who wants to hide a half-finished checkout behind a toggle.
Here's the good news. You almost certainly don't need to pay. If you build in Laravel, Pennant gives you flags for free in your own database. If you want flags bundled with analytics, PostHog's free tier covers a million flag requests a month. And three open-source platforms run on your own infrastructure for nothing.
Six tools below, with real pricing and who each one is wrong for.
Do you even need a feature flag service?
Be honest about what you're solving. For a single on/off toggle, an environment variable or a config value does the job. You start needing a real tool once you want percentage rollouts, per-user targeting, an audit trail, and the ability to flip something in production without a redeploy. That's the line. Below it, keep it simple. Above it, pick one of these rather than building your own, because every option here is free or cheap enough that rolling your own stops making sense.
Quick verdict
| Tool | Best for | Price | Rating |
|---|---|---|---|
| Laravel Pennant | Laravel apps that want flags with no SaaS | Free, first-party package | 4.6/5 |
| PostHog | Flags bundled with product analytics | Free up to 1M flag requests/mo | 4.7/5 |
| Flagsmith | Open source with remote config | Free self-host. Cloud free to 50K req, then $45/mo | 4.4/5 |
| Unleash | Privacy-focused, self-hosted at scale | Free self-host. Cloud $75/seat/mo | 4.3/5 |
| GrowthBook | Flags plus built-in A/B testing | Free self-host. Cloud free for 3 users, Pro $40/seat/mo | 4.4/5 |
| ConfigCat | Simplest managed, unlimited seats | Free (10 flags). Pro $110/mo | 4.3/5 |
| LaunchDarkly | Enterprise governance and scale | Free (1K MAU cap), then usage-based | 3.8/5 |
Laravel Pennant: flags with no extra service
If you build in Laravel, start here. Laravel Pennant is the framework's first-party feature flag package, and it's free. You install it with composer require laravel/pennant, run one migration, and you have a features table in your own database doing the work. No external service, no per-request meter, no dashboard subscription.
The API is clean. You define a flag in a service provider, check it with Feature::active('new-dashboard'), and gate Blade views with @feature. Percentage rollouts are one line with Lottery::odds(1, 10) for a 10% release, and once a user is bucketed, Pennant stores the result so they don't flicker between states. It also supports rich non-boolean values, class-based feature definitions, and route middleware.
Pricing is nothing beyond the database you already run. That's the whole point.
Who should not use Pennant: teams that need a hosted UI where non-technical people flip flags themselves. Pennant is code-first, so a product manager can't toggle a feature without you building an admin screen (Filament makes this easy, but it's still work). It's scoped to one Laravel app, and it has no built-in experimentation statistics. If you need A/B analysis or cross-service flags, look further down this list.
PostHog: flags plus analytics on one free tier
PostHog is the pick if you already want product analytics, because feature flags come bundled and the free tier is huge. You get 1 million flag requests a month free, alongside 1 million analytics events, 5,000 session recordings, and more. More than 90% of companies never pay. There's no per-seat charge either, so your whole team can log in without the bill moving.
The flags themselves are solid: percentage rollouts, user targeting, local evaluation to cut latency, and experiments with real statistical analysis sitting right next to the flag. Because it's one platform, you can ship a flag, watch the analytics, and check a session replay of the new flow without wiring three tools together. Past the free tier, flag requests cost $0.0001 each and step down sharply at volume.
PostHog is open source and self-hostable too, though that path is heavy.
Who should not use PostHog: anyone who wants only feature flags and nothing else. The platform is a suite, and its bill is built from many separate meters (events, replays, surveys, flags), so it's easy to overspend once you turn everything on. Self-hosting is not lightweight either, it needs ClickHouse, Kafka, Postgres, and Redis. If your search is strictly "cheapest flag toggle," this is more machine than you need.
Flagsmith: open source with remote config
Flagsmith is the open-source option built specifically around feature flags and remote configuration. It's BSD licensed, so you can self-host it on your own infrastructure with unlimited flags, environments, and users at zero cost. That's the strongest reason to pick it: full control and no vendor lock-in.
Beyond simple toggles, Flagsmith does remote config well. You can change app settings, copy, or limits without shipping a new build, and a single flag can carry a JSON value that controls several related settings at once. Identity management lets you target users by traits like plan, country, or signup date, and every change is logged in an audit trail.
On the managed side, the cloud free tier gives you 50,000 requests a month on a single seat, and paid plans start at $45/month. Self-hosting stays free with no request limits.
Who should not use Flagsmith: teams that want advanced governance without paying. On the self-hosted open-source build, SSO, role-based access control, and full audit logs are gated to the Enterprise tier. Its A/B testing is basic and leans on an analytics integration for real statistical work, so if experimentation is your main goal, GrowthBook or PostHog fit better. And self-hosting means you're running PostgreSQL and handling maintenance yourself.
Unleash: the established self-hosted platform
Unleash is the most mature open-source flag platform, with over 10,000 GitHub stars and 20 million-plus downloads. It's popular with privacy-conscious teams because user data never has to leave your infrastructure. You define targeting rules, gradual rollouts, and custom activation strategies, and the SDK coverage is the widest of any tool here.
Self-hosting the open-source version is free and works well for a single project. The managed cloud runs $75 per seat per month, which includes a generous 53 million API requests and adds SSO, RBAC, SOC 2, and change-approval workflows. Enterprise pricing is custom, with self-hosted Enterprise starting at a 5-seat minimum.
Who should not use Unleash: small teams watching per-seat cost. At $75 a seat, the managed cloud adds up faster than usage-based or flat-rate rivals once you have a few people. The free open-source edition is also limited to a single environment and a single project, so multi-environment setups push you toward paid or a more involved self-hosted deployment. It requires a relational database backend, and like Flagsmith, it has no built-in experimentation statistics.
GrowthBook: flags with real A/B testing built in
GrowthBook is the open-source choice when experimentation matters as much as the flag. It's MIT licensed and free to self-host with unlimited users, flags, and experiments. The cloud Starter plan is free for up to three users, also with unlimited flags and experiments, which is unusually generous.
What sets it apart is the stats. GrowthBook is warehouse-native, so it can run experiment analysis directly on your data in Snowflake, BigQuery, or Redshift without exporting events to a third party. It offers both Bayesian and frequentist engines, and the JavaScript SDK is only 9KB with local evaluation. The Pro tier at $40 per seat per month unlocks the heavier statistical methods like CUPED variance reduction and sequential testing.
Who should not use GrowthBook: solo devs who just want a kill switch. If you never plan to run a real experiment, its warehouse-native, stats-heavy design is more than you need, and simpler tools will get you shipping faster. The advanced statistics that justify GrowthBook sit behind the Pro plan, so the free tier is more of a flag tool than a full experimentation platform until you upgrade.
ConfigCat: the simplest managed option
ConfigCat wins on simplicity and predictable pricing. Its free forever plan gives you 10 flags across 2 products and 2 environments, and here's the part that matters: unlimited team members and SSO/SAML on every plan, including free. It doesn't charge per seat or per monthly active user, so the bill never surprises you.
Setup is fast, the SDK coverage is broad, and the dashboard is clean enough that non-technical teammates can manage flags without help. Paid plans are flat: Pro is $110/month for 100 flags and higher limits, Smart is $325/month for unlimited flags, and larger tiers exist above that. Everyone gets the same features, only the limits change.
Who should not use ConfigCat: teams that will blow past 10 flags quickly and don't want the jump to $110/month. The free tier is genuinely useful but its flag cap is low, so a busy app outgrows it faster than PostHog's request-based free tier. And if deep experimentation statistics are your reason for buying, ConfigCat supports experimentation workflows through analytics integrations but isn't trying to be a full A/B testing suite.
How do you choose the right one?
Match the tool to your stack and your actual need.
- You build in Laravel: Laravel Pennant. Free, first-party, in your own database. Nothing to host.
- You want flags plus analytics for free: PostHog. A million flag requests a month, no per-seat cost.
- You want open source with remote config: Flagsmith. Self-host free, or a cheap cloud entry.
- You care about data residency and self-hosting: Unleash. The most established option, privacy-first.
- You want to A/B test, not just toggle: GrowthBook. Flags and real experimentation in one place.
- You want the simplest managed dashboard: ConfigCat. Unlimited seats and SSO even on the free plan.
The verdict
For Laravel builders, Laravel Pennant is the obvious default. It's free, it lives in your own database, and it does everything a solo founder needs for safe rollouts. If you're not on Laravel, PostHog is the best all-round free pick because the flag tier is huge and you get analytics in the same tool. Want to genuinely experiment rather than just toggle? GrowthBook. Want the least fuss and a dashboard your teammates can use? ConfigCat.
LaunchDarkly is a genuinely strong platform, and if you're a funded team that needs enterprise governance, SSO/SCIM, and release automation, it earns its price. But that's not most people reading this. The honest takeaway is that feature flags are close to a solved problem at the indie scale, and you can have them for free.
Feature flags pair naturally with the rest of your release safety net. If you're setting up error tracking alongside them, our Sentry alternatives guide covers the cheap options, and for deployment our take on Laravel Forge vs Ploi vs Coolify fits the same trunk-based workflow. If you go the PostHog route, our PostHog vs Plausible vs Fathom vs Mixpanel and Hotjar vs Microsoft Clarity vs PostHog comparisons go deeper on the analytics side.
Top comments (0)