Prisma [NextJS]: Many-to-many query is very slow

Question

I have a nextJS project and I am using Prisma as an ORM.

The concerning database-tables look like this:

model cars_archive {
  id               Int       @id @default(autoincrement())
  ...
  inactivestamp    DateTime? @db.DateTime(0)

  @@index([brand], map: "brand")
  @@index([model], map: "model")
}

model extras {
  id          Int           @id @default(autoincrement()) @db.MediumInt
  value       String        @unique(map: "value") @db.VarChar(255)
  extras_cars extras_cars[]
}

model extras_cars {
  car_id   Int
  extra_id Int    @db.MediumInt
  extras   extras @relation(fields: [extra_id], references: [id], onUpdate: Restrict, map: "extras_cars_ibfk_1")
  cars     cars   @relation(fields: [car_id], references: [id], onUpdate: Restrict, map: "extras_cars_ibfk_2")

  @@id([car_id, extra_id])
  @@index([car_id, extra_id], map: "car_id")
  @@index([extra_id], map: "extra_id")
}

I need a query that fetches all cars that have certain extras. Something like this: ['Cruise Control', 'ABS',...]

To do that I created the following query-statement:

const cars = await prisma.cars.findMany({
        where: {
            extras_cars: {
                every: {
                    extras: {
                        value: {
                            in: ['Cruise Control','ABS']
                        }
                    }
                }
            }
        }
    })

The thing is the query yields the desired result, but takes forever to return the list of cars (15 - 20 seconds).

I have tried the same thing with a simple SQL-Query which looks like this: SELECT COUNT(*) FROM cars JOIN extras_cars ec on cars.id = ec.car_id JOIN extras e on e.id = ec.extra_id WHERE e.value='Cruise Control' This query just takes about 1 second.

When looking at the debug-logs from Prisma it shows a really long SQL-query like this, which just doesn't make any sense to me: SELECT COUNT(*) FROM (SELECT carsDB.cars.idFROMcarsDB.cars WHERE (carsDB.cars.id) NOT IN (SELECT t0.idFROMcarsDB.carsASt0INNER JOINcarsDB.extras_carsASj0 ON (j0.car_id) = (t0.id) WHERE ((NOT (j0.car_id,j0.extra_id) IN (SELECT t1.car_id, t1.extra_idFROMcarsDB.extras_carsASt1INNER JOINcarsDB.extrasASj1 ON (j1.id) = (t1.extra_id) WHERE (j1.value= ? ANDt1.car_idIS NOT NULL ANDt1.extra_idIS NOT NULL))) ANDt0.idIS NOT NULL)) /* traceparent=00-00-00-00 */) ASsub /* traceparent=00-00-00-00 */

Has somebody ever had a similar issue or has an idea of how to speed up the query?

The documentation of Prisma doesn't yield any useful information.

BTW: My database is a mariadb.

Thanks in advance.