Today, we are issuing a 7.4.2 patch release focused on bug fixes and quality improvements.

🛠 Fixes

Prisma Client

• Fix a case-insensitive IN and NOT IN filter regression (https://github.com/prisma/prisma/pull/29243)
• Fix a query plan mutation issue that resulted in broken cursor queries (https://github.com/prisma/prisma/pull/29262)
• Fix an array parameter wrapping issue in push operations (https://github.com/prisma/prisma-engines/pull/5784)
• Fix Uint8Array serialization in nested JSON fields (https://github.com/prisma/prisma/pull/29268)
• Fix an issue with MySQL joins that relied on non-strict equality (https://github.com/prisma/prisma/pull/29251)

Driver Adapters

• @prisma/adapter-mariadb: Update text column detection to check for a binary collation (https://github.com/prisma/prisma/pull/29238)
• @prisma/adapter-mariadb: Correct relationJoins compatibility check for MariaDB 8.x versions (https://github.com/prisma/prisma/pull/29246)

Schema Engine

• Fix partial index predicate comparison on PostgreSQL and MSSQL (https://github.com/prisma/prisma-engines/pull/5780)

🙏 Huge thanks to our community

Many of the fixes in this release were contributed by our amazing community members. We're grateful for your continued support and contributions that help make Prisma better for everyone!

Today, we are issuing a 7.4.1 patch release focused on bug fixes and quality improvements.

🛠 Fixes

Prisma Client

• Fix cursor-based pagination regression with parameterised values (https://github.com/prisma/prisma/pull/29184)
• Preserve Prisma.skip through query extension argument cloning (https://github.com/prisma/prisma/pull/29198)
• Enable batching of multiple queries inside interactive transactions (https://github.com/prisma/prisma/pull/25571)
• Add missing JSON value deserialization for JSONB parameter fields (https://github.com/prisma/prisma/pull/29182)
• Apply result extensions correctly for nested and fluent relations (https://github.com/prisma/prisma/pull/29218)
• Allow missing config datasource URL and validate only when needed (https://github.com/prisma/prisma-engines/pull/5777)

Driver Adapters

• @prisma/adapter-ppg: Handle null values in type parsers for nullable columns (https://github.com/prisma/prisma/pull/29192)

Prisma Schema Language

• Support where argument on field-level @unique for partial indexes (https://github.com/prisma/prisma-engines/pull/5774)
• Add object expression and object member support to schema reformatter (https://github.com/prisma/prisma-engines/pull/5776)

🙏 Huge thanks to our community

Many of the fixes in this release were contributed by our amazing community members. We're grateful for your continued support and contributions that help make Prisma better for everyone!

Today, we are excited to share the 7.4.0 stable release 🎉

🌟 Star this repo for notifications about new releases, bug fixes & features — or follow us on X!

Highlights

ORM

Caching in Prisma Client

Today’s release is a big one, as we introduce a new caching layer into Prisma ORM. But why the need for a caching layer?

In Prisma 7, the query compiler runs as a WebAssembly module directly on the JavaScript main thread. While this simplified the architecture by eliminating the separate engine process, it introduced a trade-off: every query now synchronously blocks the event loop during compilation.

For individual queries, compilation takes between 0.1ms and 1ms, which is barely noticeable in isolation. But under high concurrency this overhead adds up and creates event loop contention that affects overall application throughput.

For instance, say we have a query that is run over and over, but is a similar shape:

// These two queries have the same shape:
const alice = await prisma.user.findUnique({ where: { email: '[email protected]' } })
const bob = await prisma.user.findUnique({ where: { email: '[email protected]' } })

Prior to v7.4.0, this would be reevaluated ever time the query is run. Now, Prisma Client will extract the user-provided values and replaces them with typed placeholders, producing a normalized query shape:

prisma.user.findUnique({ where: { email: %1 } })   // cache key
                                         ↑
                              %1 = '[email protected]'  (or '[email protected]')

This normalized shape is used as a cache key. On the first call, the query is compiled as usual and the resulting plan is stored in an LRU cache. On every subsequent call with the same query shape, regardless of the actual values, the cached plan is reused instantly without invoking the compiler.

We have more details on the impact of this change and some deep dives into Prisma architecture in an upcoming blog post!

Partial Indexes (Filtered Indexes) Support

We're excited to announce Partial Indexes support in Prisma! This powerful community-contributed feature allows you to create indexes that only include rows matching specific conditions, significantly reducing index size and improving query performance.

Partial indexes are available behind the partialIndexes preview feature for PostgreSQL, SQLite, SQL Server, and CockroachDB, with full migration and introspection support.

Basic usage

Enable the preview feature in your schema:

generator client {
  provider        = "prisma-client-js"
  previewFeatures = ["partialIndexes"]
}

Raw SQL syntax

For maximum flexibility, use the raw() function with database-specific predicates:

model User {
  id       Int     @id
  email    String
  status   String

  @@unique([email], where: raw("status = 'active'"))
  @@index([email], where: raw("deletedAt IS NULL"))
}

Type-safe object syntax

For better type safety, use the object literal syntax for simple conditions:

model Post {
  id        Int      @id
  title     String
  published Boolean

  @@index([title], where: { published: true })
  @@unique([title], where: { published: { not: false } })
}

Bug Fixes

Most of these fixes are community contributions - thank you to our amazing contributors!

• prisma/prisma-engines#5767: Fixed an issue with PostgreSQL migration scripts that prevented usage of CREATE INDEX CONCURRENTLY in migrations
• prisma/prisma-engines#5752: Fixed BigInt precision loss in JSON aggregation for MySQL and CockroachDB by casting BigInt values to text (from community member polaz)
• prisma/prisma-engines#5750: Fixed connection failures with non-ASCII database names by properly URL-decoding database names in connection strings
• #29155: Fixed silent transaction commit errors in PlanetScale adapter by ensuring COMMIT failures are properly propagated
• #29141: Resolved race condition errors (EREQINPROG) in SQL Server adapter by serializing commit/rollback operations using mutex synchronization
• #29158: Fixed MSSQL connection string parsing to properly handle curly brace escaping for passwords containing special characters

Open roles at Prisma

Interested in joining Prisma? We’re growing and have several exciting opportunities across the company for developers who are passionate about building with Prisma. Explore our open positions on our Careers page and find the role that’s right for you.

Enterprise support

Thousands of teams use Prisma and many of them already tap into our Enterprise & Agency Support Program for hands-on help with everything from schema integrations and performance tuning to security and compliance.

With this program you also get priority issue triage and bug fixes, expert scalability advice, and custom training so that your Prisma-powered apps stay rock-solid at any scale. Learn more or join: https://prisma.io/enterprise.