Jotai v2.20: Rework Store Building Blocks for High-Throughput Performance and Sets the Stage for v3

Jotai v2.20.0 has been released with a focus on improved performance in high-throughput scenarios. The release centers on reworking internal store building blocks and fixing earlier performance regressions. Everyday application code keeps the same API surface. The report is from InfoQ by Daniel Curtis.

Technically, the work targets the library’s store building blocks rather than the public hooks and atoms most apps already use. That keeps the day-to-day API stable while the internals are adjusted for higher throughput. The same pass also rolls back earlier regressions so high-update workloads do not pay an unexpected cost. Deprecations aimed at library authors sit alongside this, marking the first concrete steps toward Jotai v3.

For engineers and builders, the main benefit is that most teams can take the performance work without rewriting atom trees or store wiring. Apps that push frequent updates through the store stand to gain the most from the high-throughput focus. Library authors who extend or wrap Jotai need to treat the deprecations as real signals, not optional cleanup, because those APIs are being reshaped for the next major line.

In market and competitive terms, Jotai stays in the atomic-state space where update volume and store internals often decide whether a library feels light or heavy under load. Shipping a performance-focused 2.x cut while holding the everyday API steady is a conservative path: users get throughput work now, and v3 groundwork proceeds without a forced break. Deprecations for library authors draw a clean line between app consumers and package authors who sit on the extension surface.

The practical takeaway is to upgrade on the performance and regression fixes if you already depend on Jotai, and keep the everyday API assumptions intact. If you publish a library on top of Jotai, inventory usage of the deprecated building blocks and plan replacements before v3 hardens them. What to watch next is how those deprecations map into the v3 store and authoring model, and whether high-throughput gains hold once that major lands.