Harness FME (formerly Split) Review (2026)

What Harness FME actually is

Harness FME is what used to be Split. Split was one of only two flag platforms that took experimentation seriously (LaunchDarkly being the other), pairing feature flags with statistically rigorous A/B analysis. In June 2024, Harness - a well-funded software-delivery platform founded in 2017 by Jyoti Bansal - completed its acquisition of Split and rebranded it Feature Management & Experimentation (FME), a module inside the wider Harness developer platform.

The important framing: this is a rebrand, not a sunset. The product is actively developed, the engine works, and the experimentation capability that made Split notable is intact. What changed is how you buy it and what it’s part of. Harness raised a $240M Series E in December 2025 led by Goldman Sachs at a $5.5B valuation, so the parent is solidly resourced.

The strength: experimentation-grade feature management

This is why Split mattered and why FME still does. Most flag tools bolt a rollout percentage onto a toggle and call it experimentation. Split built statistically rigorous experiment analysis and metric-impact measurement driven by feature flags - real testing, not a progress bar. That carries into FME, now wrapped in Harness enterprise controls for experiments and rollouts and a unified UI.

In ambition it’s comparable to GrowthBook. The difference is philosophy: GrowthBook is open source and warehouse-native; FME is proprietary, hosted and enterprise-priced. If you want serious experimentation inside a managed enterprise platform and don’t care about self-hosting, FME is a real contender.

Pricing: the standalone option is gone

Here’s the change that matters most for buyers. Standalone Split pricing no longer exists. The old split.io/pricing page now 301-redirects to harness.io/pricing. FME is listed as a module under the Harness Enterprise plan on a contact-sales basis, with no per-tier numbers and no published MAU quota.

Plan	Price	Notes
FME Free Plan	$0	Required signup step; exact quota and card requirement not published
Enterprise (FME module)	Contact sales	MAU-based, part of the Harness platform bundle

The historical model was MAU-based, and that carries into FME. Third-party market data suggests real deals land across a wide five-to-six-figure annual range depending on MAU, flag count and experimentation modules - but that’s rough market intelligence, not a quotable figure, so treat it as directional only.

The gotcha: absorption into the Harness platform

The real risk here isn’t technical - it’s structural. Split has been absorbed into a larger DevOps suite, and you can no longer buy it on its own. For a small team that just wanted Split’s flags and experiments as a standalone product, that path is closed. You’re now buying into, or migrating into, the Harness platform.

For existing Split customers the migration itself was handled well: the back-end is unchanged, so no application code, SDK, Proxy, Synchronizer or Evaluator changes were required. The friction is around the edges. SSO must be reconfigured as a new Harness SSO connection, set up at least a week ahead. Five Split admin API endpoints changed, so any automation scripts need updating. And legacy Split permissions must be reconciled with Harness RBAC. Nothing catastrophic, but it’s work, and there’s an underlying repricing risk as the standalone deal becomes a platform-bundle deal at renewal.

Should you use it?

Use Harness FME if you’re an enterprise that wants experimentation-grade feature management inside a managed platform, you’re already using or open to adopting the wider Harness suite, and self-hosting isn’t a requirement. The experimentation depth is genuine and the parent company is well-funded.

Don’t use Harness FME if you wanted Split as a lightweight standalone product (that’s gone), you need self-hosting or open source (look at Unleash or GrowthBook), or you’re a small team that doesn’t want to buy into an enterprise DevOps platform to get feature flags. The contact-sales, MAU-based model isn’t built for you.

Bottom line: the engine that made Split worth using is intact, and the migration didn’t break anyone’s code. But Split as a standalone buy no longer exists - it’s a Harness module now, priced and packaged as part of a larger platform. Evaluate it as a Harness commitment, not a point tool, because that’s what it has become.

---

Pricing and features verified against harness.io on 23 July 2026. Feature-flag pricing changes often - we re-verify regularly. FME pricing is contact-sales; any contract figures referenced are third-party market intelligence and labelled as such.