Today, we’re releasing Effect 4.0.
It’s the most ambitious release we’ve ever shipped. We rebuilt Effect from the ground up. It’s faster, uses a fraction of the memory, and tree-shakes to far smaller bundles.
- Smaller bundle size
- 5×
- Concurrent task throughput
- 6.4×
- Less memory per fiber
- 86%
Minimal program size lower is better
- 3.x
- 35.6 kB
- 4.x
- 7.1 kB
Tasks per second higher is better
- 3.x
- 0.71M
- 4.x
- 4.57M
Heap for 50,000 fibers lower is better
- 3.x
- 157.5 MB
- 4.x
- 21.8 MB
Bundles are built from identical source, minified and gzipped. Runtime figures are medians of nine fresh-process runs.
One ecosystem, zero dependencies
Many previously separate packages now live in effect itself. The packages around it share a single version and release in lockstep.
The core effect package has zero runtime dependencies. We control every line we ship, so there are no third-party dependency chains and less exposure to supply chain attacks.
From functions to clusters
Effect started with a much smaller scope. Today, it covers the full range, from a single function to a distributed system.
People came for the programming model: an effect system with built-in typed errors, dependency injection, resource management, structured concurrency, and observability. The standard library and platform integrations grew around it. Then came higher-level primitives for distributed systems, including durable workflows and clustering.
Every layer is built on that same model, with the same guarantees all the way up.
Mainstream adoption
Effect now runs in production at companies of all sizes, including large enterprises.
- Downloads in the week of September 21, 2026
- 43.9M
- Growth since 3.x
- 179×
4.x adoption already outsizes 3.x last 7 days
- 4.x
- 56%
- 3.x
- 44%
Projects with communities of their own have grown around it: Alchemy for cloud infrastructure, and Foldkit for the frontend.
If your team runs Effect in production, let’s talk. Tell us what you’re building and how we can help.
Long-term support
Teams are betting their products on Effect, and they deserve commitments they can plan around. Starting with this release, every major version of Effect has a long-term support policy.
For Effect 4.x, that means:
- Bug fixes until September 2029, or one year after 5.0 ships, whichever is later
- Security fixes until September 2029, or two years after 5.0 ships, whichever is later
2026 2028 2030 2032
Sep 20275.0 ships
5.0 ships
5.0 ships
Support lasts at least three years. If 5.0 ships late enough, bug and security fixes run past that minimum.
We’ll share the full policy and our release process in a follow-up post.
What’s next
Our first priority is stabilizing more of the ecosystem. Some of the newer modules are still tagged @stability unstable or @stability experimental, which means they may change in minor or patch releases. We’ll use production feedback to refine them and promote them to stable.
Beyond that, we’ll continue expanding Effect into a single programming model for your entire application, with broader native platform support and more higher-level primitives.
Over to you
Migrating? Start with the migration guide. Hand it to your coding agent, and it should do most of the work on its own.
Curious? Check out our documentation or ask your agent why you should be building with Effect.
Thank you to everyone who contributed, tested releases, reported bugs, and helped others get started. Effect is better because of you.
Spread the word and celebrate with us on X, Bluesky, and Discord.
Thank you for your attention to this matter.