Fix Slow Prisma Queries With Direct SQL Execution

Love Prisma's DX but need lower read overhead on supported query shapes? Current benchmark JSON shows average speedups across Prisma v6, v7, and v8 for supported PostgreSQL and SQLite reads; individual query shapes vary.

Speed Up Prisma Now

Faster Supported Reads
No Query Changes
Current Benchmark JSON
PostgreSQL & SQLite

Your Prisma calluser.findMany

const users = await prisma.user.findMany({
  where: { status: 'ACTIVE' },
  include: { posts: true },
})

SQL that prisma-sql runs

SELECT user_t.id, user_t.email, user_t.status,  COALESCE((    SELECT COALESCE(json_agg(json_build_object(      'id', posts_0.id,      'title', posts_0.title,      'authorId', posts_0."authorId"    )), '[]'::json)    FROM "public"."Post" posts_0    WHERE posts_0."authorId" = user_t.id  ), '[]'::json) AS postsFROM "public"."User" user_tWHERE user_t.status = $1

params ["ACTIVE"]1 statement

SELECT user_t.id, user_t.email, user_t.status,  COALESCE((    SELECT COALESCE(json_group_array(row), json('[]'))    FROM (      SELECT json_object(        'id', posts_0.id,        'title', posts_0.title,        'authorId', posts_0."authorId"      ) AS row      FROM "Post" posts_0      WHERE posts_0."authorId" = user_t.id    ) AS posts_row_1  ), json('[]')) AS postsFROM "User" user_tWHERE user_t.status = ?

params ["ACTIVE"]1 statement

Output of createToSQL for a two-model schema, line breaks added.

Prisma Coverage: v6–v8; PostgreSQL and SQLite benchmark matrix
Query Changes: 0; Keep your existing Prisma calls
Benchmark Files: 6; Prisma v6/v7/v8 × PG/SQLite
Type Safety: 100%; Same Prisma API

Why Is Prisma Slow?

Understanding Prisma's evolution and performance characteristics helps explain why this extension exists.

The Prisma Journey: DX First, Performance Matters at Scale

💡 Why This Extension Exists

Prisma made the right architectural choices for its goals: type safety, developer experience, and cross-database behavior. But those choices create overhead that's noticeable at scale. This extension doesn't replace Prisma—it optimizes reads for teams that want Prisma's DX plus faster execution where it matters.

🎯 2019: The Birth of Modern Prisma
Prisma 2 launched in 2019, changing the TypeScript ORM landscape with type-safe database access and generated types from your schema. Prisma introduced an engine-based architecture to translate queries, validate them, and provide strong guarantees that were hard to achieve with traditional JavaScript ORMs.
⚡ 2020–2022: Rapid Growth & Feature Expansion
Prisma added powerful features like nested writes, transactions, and middleware. The feature set expanded, but every query still paid an architectural cost: queries are represented, validated, executed, and results are shaped to match the Prisma API.
📊 2023: Overhead Becomes Noticeable at Scale
As more teams deployed Prisma in high-traffic workloads, the fixed per-query overhead became measurable. This is most visible on read-heavy endpoints, analytics, aggregations, and large result sets. The overhead is not a bug; it's the cost of Prisma’s guarantees and API behavior.
🚀 2024–2025: Prisma Performance Work Continues
Prisma shipped major updates focused on performance and engine changes. Even with improvements, there is still an unavoidable cost to parsing, validating, planning, and shaping results compared to executing raw SQL directly.
🎯 2026: prisma-sql Extension Released
This extension focuses on read performance. It bypasses Prisma’s read execution path for findMany, findFirst, findUnique, count, aggregate, and groupBy, while keeping Prisma for writes, migrations, schema management, and type generation. Validate compatibility with your Prisma version before rollout.

Query Translation Layer

Prisma translates your query inputs into database-specific SQL. This enables cross-database behavior and Prisma's API semantics, but adds processing time before the database sees the SQL.

Validation & Type Guarantees

Prisma validates queries against the schema and enforces API-level guarantees. These safeguards prevent classes of bugs, but they also add overhead to each query.

Result Shaping

Results are shaped to match Prisma's API behavior. This is great for DX and consistency, but it adds latency, especially on large result sets and complex includes.

This extension complements Prisma by offering a faster path for supported read queries. You keep Prisma for schema, migrations, writes, and generated types while generated SQL handles selected hot reads through database-native clients.

How Much Faster? Real Benchmarks

Latest benchmark JSON comparison across Prisma v6, v7, v8, Drizzle ORM, and prisma-sql

PostgreSQL

95 query cases per run

Prisma v6 (6.19.3)1.56× vs Drizzle
1.99×
Prisma v6 (6.19.3): 1.99× average speedup versus Prisma
95 measured query cases, average vs PrismaSep 30, 2026
Prisma v7 (7.10.0)1.51× vs Drizzle
1.47×
Prisma v7 (7.10.0): 1.47× average speedup versus Prisma
95 measured query cases, average vs PrismaSep 30, 2026
Prisma v8 (8.1.0-dev.1)1.63× vs Drizzle
1.54×
Prisma v8 (8.1.0-dev.1): 1.54× average speedup versus Prisma
95 measured query cases, average vs PrismaSep 30, 2026

SQLite

84 query cases per run

Prisma v6 (6.19.3)1.79× vs Drizzle
4.80×
Prisma v6 (6.19.3): 4.80× average speedup versus Prisma
84 measured query cases, average vs PrismaSep 30, 2026
Prisma v7 (7.10.0)1.84× vs Drizzle
2.29×
Prisma v7 (7.10.0): 2.29× average speedup versus Prisma
84 measured query cases, average vs PrismaSep 30, 2026
Prisma v8 (8.1.0-dev.1)1.63× vs Drizzle
2.10×
Prisma v8 (8.1.0-dev.1): 2.10× average speedup versus Prisma
84 measured query cases, average vs PrismaSep 30, 2026

Generated at build time from the latest benchmark-results/*-latest.json files. Bars share one scale across both databases; amber marks the best run per database. Individual query shapes vary.Per-query table View complete benchmark data

Detailed Benchmark Results

Statistical comparison with Welch's t-test and a 1ms practical significance threshold

These results reflect runtime mode, where queries are converted to SQL on each request. In prerendered mode, SQL is generated at build time — runtime executes raw parameterized queries with zero conversion overhead.

PostgreSQLPrisma v6 (6.19.3)

Avg 1.99x vs PrismaAvg 1.56x vs Drizzle35/95 significant wins60 within noise floor

Sep 30, 2026

Test	Prisma	prisma-sql	Drizzle	Speedup	Sig.	CV%	n
findMany basic	0.529ms±0.064	0.231ms±0.234	0.271ms	~1.00x	≈	198.0%	50
findMany where =	0.501ms±0.0069	0.208ms±0.0094	0.347ms	~1.00x	≈	15.6%	50
findMany where >=	16.51ms±0.267	5.67ms±0.338	7.43ms	2.91x1.31x D	***	21.0%	50
findMany where IN	0.657ms±0.142	0.292ms±0.017	0.492ms	~1.00x	≈	20.3%	50
findMany where null	0.280ms±0.020	0.167ms±0.012	0.177ms	~1.00x	≈	25.7%	50
findMany ILIKE	0.297ms±0.0097	0.168ms±0.0089	0.278ms	~1.00x	≈	18.6%	50
findMany AND	2.72ms±0.161	1.11ms±0.111	1.45ms	2.45x1.31x D	***	31.5%	50
findMany OR	13.98ms±0.286	4.69ms±0.221	5.88ms	2.98x1.25x D	***	17.0%	50
findMany NOT	0.639ms±0.013	0.274ms±0.0091	0.478ms	~1.00x	≈	12.2%	50
findMany orderBy	4.50ms±0.219	1.48ms±0.075	1.76ms	3.03x1.18x D	***	18.5%	50
findMany pagination	0.342ms±0.032	0.171ms±0.026	0.295ms	~1.00x	≈	43.2%	50
findMany select	0.294ms±0.0094	0.181ms±0.0072	0.176ms	~1.00x	≈	14.0%	50
findMany relation some	1.69ms±0.098	0.999ms±0.064	—	~1.00x	≈	22.1%	50
findMany relation every	2.17ms±0.096	1.47ms±0.078	—	~1.00x	≈	18.9%	50
findMany relation none	26.02ms±0.517	10.06ms±0.511	—	2.59x	***	18.0%	50
findMany nested relation	1.76ms±0.076	1.24ms±0.060	—	~1.00x	≈	18.0%	50
findMany complex	2.94ms±0.109	1.46ms±0.068	1.73ms	2.01x1.18x D	***	17.2%	50
findFirst	0.378ms±0.032	0.175ms±0.014	0.350ms	~1.00x	≈	28.0%	50
findFirst skip	0.328ms±0.107	0.147ms±0.0064	0.301ms	~1.00x	≈	15.2%	50
findUnique id	0.232ms±0.0097	0.102ms±0.0055	0.207ms	~1.00x	≈	18.6%	50
findUnique email	0.210ms±0.0072	0.090ms±0.0044	0.179ms	~1.00x	≈	17.1%	50
count	0.109ms±0.0033	0.063ms±0.0030	0.064ms	~1.00x	≈	16.2%	50
count where	0.469ms±0.015	0.247ms±0.016	0.287ms	~1.00x	≈	22.2%	50
aggregate count	0.215ms±0.0078	0.155ms±0.0042	—	~1.00x	≈	9.2%	50
aggregate sum/avg	0.305ms±0.014	0.261ms±0.0053	—	~1.00x	≈	7.1%	50
aggregate where	0.446ms±0.0100	0.239ms±0.0058	—	~1.00x	≈	8.7%	50
aggregate min/max	0.305ms±0.015	0.268ms±0.010	—	~1.00x	≈	13.2%	50
aggregate complete	0.381ms±0.0061	0.292ms±0.0072	—	~1.00x	≈	8.8%	50
groupBy	0.419ms±0.024	0.344ms±0.013	—	~1.00x	≈	12.8%	50
groupBy count	0.422ms±0.0100	0.358ms±0.0050	—	~1.00x	≈	4.8%	50
groupBy multi	0.544ms±0.013	0.442ms±0.052	—	~1.00x	≈	38.9%	50
groupBy having	0.505ms±0.011	0.375ms±0.0086	—	~1.00x	≈	8.2%	50
groupBy + where	0.550ms±0.011	0.294ms±0.0064	—	~1.00x	≈	7.8%	50
_count via include	0.679ms±0.013	0.511ms±0.013	—	~1.00x	≈	8.9%	50
_count via include + relations	1.55ms±0.037	0.666ms±0.017	—	~1.00x	≈	9.3%	50
groupBy aggregates	0.557ms±0.0097	0.465ms±0.0091	—	~1.00x	≈	6.9%	50
groupBy min/max	0.531ms±0.0080	0.441ms±0.0053	—	~1.00x	≈	4.2%	50
include list + nested to-one	1.46ms±0.029	0.337ms±0.032	—	4.32x	***	30.2%	50
include list + nullable to-one	2.59ms±0.081	1.65ms±0.056	—	~1.00x	≈	11.9%	50
include posts	4.25ms±0.216	2.35ms±0.120	5.77ms	1.81x2.45x D	***	18.9%	50
include profile	0.635ms±0.871	0.540ms±0.241	0.648ms	~1.00x	≈	104.7%	50
include 3 levels	4.48ms±0.148	4.03ms±0.147	4.58ms	~1.00x	≈	13.6%	50
include 4 levels	5.61ms±0.182	3.39ms±0.132	4.65ms	1.65x1.37x D	***	13.9%	50
include + where	4.04ms±0.176	2.88ms±0.399	5.51ms	~1.00x	≈	46.7%	50
include + select nested	2.52ms±0.139	1.96ms±0.185	4.42ms	~1.00x	≈	31.8%	50
findMany startsWith	0.270ms±0.028	0.140ms±0.014	0.314ms	~1.00x	≈	35.4%	50
findMany endsWith	0.563ms±0.015	0.216ms±0.0069	0.369ms	~1.00x	≈	11.1%	50
findMany NOT contains	0.627ms±0.022	0.250ms±0.104	0.447ms	~1.00x	≈	121.8%	50
findMany LIKE	0.245ms±0.0094	0.111ms±0.0069	0.250ms	~1.00x	≈	21.9%	50
findMany <	25.55ms±0.398	7.55ms±0.292	10.60ms	3.39x1.40x D	***	13.4%	50
findMany <=	26.43ms±0.371	7.78ms±0.220	10.97ms	3.40x1.41x D	***	10.2%	50
findMany >	15.55ms±0.599	5.21ms±0.356	6.78ms	2.99x1.30x D	***	24.2%	50
findMany NOT IN	0.644ms±0.013	0.282ms±0.011	0.487ms	~1.00x	≈	13.4%	50
findMany isNot null	0.642ms±0.014	0.336ms±0.017	0.362ms	~1.00x	≈	18.6%	50
orderBy multi-field	4.99ms±0.190	1.55ms±0.076	1.88ms	3.22x1.21x D	***	17.0%	50
orderBy relation to-one	4.10ms±0.135	3.08ms±0.090	—	1.33x	***	10.2%	50
orderBy relation + scalar	5.02ms±0.258	3.63ms±0.205	—	1.39x	***	20.0%	50
orderBy nullable relation	4.08ms±0.300	2.94ms±0.181	—	1.39x	***	21.1%	50
orderBy deep relation	5.43ms±0.186	4.04ms±0.174	—	1.34x	***	14.9%	50
orderBy deep relation + scalar	6.34ms±0.135	4.62ms±0.100	—	1.37x	***	7.8%	50
distinct status	8.84ms±0.376	1.56ms±0.181	—	5.66x	***	36.9%	50
distinct multi	12.04ms±0.171	2.50ms±0.073	—	4.81x	***	10.3%	50
cursor composite tuple	1.61ms±0.025	1.13ms±0.017	—	~1.00x	≈	5.3%	50
cursor composite desc	1.75ms±0.037	1.31ms±0.021	—	~1.00x	≈	5.8%	50
cursor prefix of orderBy	2.70ms±0.188	1.66ms±0.082	—	1.63x	***	18.1%	50
select + include	1.60ms±0.101	0.309ms±0.021	0.568ms	5.16x1.84x D	***	24.6%	50
_count relation	1.74ms±0.341	1.67ms±0.781	—	~1.00x	≈	123.8%	50
_count multi-relation	0.257ms±0.029	0.218ms±0.045	—	~1.00x	≈	62.4%	50
_count inside include	2.32ms±0.126	1.55ms±0.124	—	~1.00x	≈	32.0%	50
_count inside nested select	4.64ms±0.202	3.69ms±0.266	—	~1.00x	≈	24.3%	50
_count deep include	5.49ms±0.247	3.98ms±0.158	—	1.38x	***	14.1%	50
ILIKE special chars	0.425ms±0.033	0.193ms±0.022	—	~1.00x	≈	36.5%	50
LIKE case sensitive	0.263ms±0.010	0.110ms±0.0078	—	~1.00x	≈	24.0%	50
findMany Date range	0.424ms±0.012	0.190ms±0.012	0.321ms	~1.00x	≈	21.1%	50
count Date range	1.14ms±0.047	0.925ms±0.052	0.995ms	~1.00x	≈	19.8%	50
findMany Date gte	0.403ms±0.025	0.178ms±0.0058	0.303ms	~1.00x	≈	11.6%	50
depth-1 low-fan	0.566ms±0.021	0.447ms±0.102	—	~1.00x	≈	73.7%	50
depth-1 mid-fan	4.21ms±0.110	2.34ms±0.178	—	1.80x	***	26.3%	50
depth-1 high-fan	4.16ms±0.122	2.25ms±0.130	—	1.85x	***	20.4%	50
depth-1 wide	3.11ms±0.114	1.78ms±0.166	—	1.75x	***	31.3%	50
depth-1 unbound	39.65ms±3.78	9.43ms±0.884	—	4.21x	***	29.1%	50
depth-2	4.78ms±0.153	3.54ms±0.187	—	1.35x	***	18.6%	50
depth-2 paginated Project→tasks	3.16ms±0.149	2.68ms±0.161	—	~1.00x	≈	20.6%	50
depth-2 high-fan	3.84ms±0.185	2.86ms±0.229	—	~1.00x	≈	28.7%	50
depth-2 wide	4.90ms±0.176	3.08ms±0.176	—	1.59x	***	20.6%	50
depth-2 unbound	65.60ms±1.52	17.60ms±0.479	—	3.73x	***	4.4%	10
depth-3 unbound	15.32ms±0.281	8.38ms±0.354	—	1.83x	***	15.8%	50
depth-3 paginated	2.44ms±0.156	2.34ms±0.253	—	~1.00x	≈	32.7%	50
depth-4 unbound	9.96ms±0.268	6.14ms±0.266	—	1.62x	***	15.7%	50
depth-4 paginated	3.42ms±0.161	2.42ms±0.175	—	~1.00x	≈	23.6%	50
findFirst depth-2	2.91ms±0.114	2.30ms±0.173	—	~1.00x	≈	25.9%	50
findUnique depth-2 Project→tasks→comment	3.27ms±0.113	2.16ms±0.173	—	1.52x	***	27.5%	50
complex nested select	4.51ms±0.428	3.01ms±0.777	—	~1.00x	≈	65.7%	50
ultra deep query	15.94ms±0.546	5.83ms±0.361	—	2.73x	***	22.0%	50
transaction: (3 operations)	3.00ms±0.275	1.41ms±0.246	—	2.14x	***	53.2%	50

*** p < 0.001** p < 0.01* p < 0.05ns not significant≈ within 1ms noise floor± 95% CICV% > 15% = high variance

Runtime mode benchmarked. Prerendered mode skips query-to-SQL conversion entirely at runtime — expect lower latency than shown here, especially for complex queries where conversion cost is proportionally higher.

PostgreSQLPrisma v7 (7.10.0)

Avg 1.47x vs PrismaAvg 1.51x vs Drizzle18/95 significant wins77 within noise floor

Sep 30, 2026

Test	Prisma	prisma-sql	Drizzle	Speedup	Sig.	CV%	n
findMany basic	0.274ms±0.012	0.223ms±0.072	0.252ms	~1.00x	≈	97.3%	50
findMany where =	0.212ms±0.0047	0.158ms±0.0044	0.244ms	~1.00x	≈	9.6%	50
findMany where >=	7.36ms±0.258	4.63ms±0.213	6.57ms	1.59x1.42x D	***	16.5%	50
findMany where IN	0.372ms±0.027	0.231ms±0.015	0.410ms	~1.00x	≈	21.1%	50
findMany where null	0.111ms±0.084	0.091ms±0.0072	0.106ms	~1.00x	≈	26.6%	50
findMany ILIKE	0.120ms±0.0036	0.088ms±0.0025	0.154ms	~1.00x	≈	10.0%	50
findMany AND	0.968ms±0.013	0.540ms±0.093	0.705ms	~1.00x	≈	56.3%	50
findMany OR	7.35ms±0.343	3.95ms±0.218	5.27ms	1.86x1.33x D	***	19.1%	50
findMany NOT	0.361ms±0.022	0.231ms±0.121	0.411ms	~1.00x	≈	149.0%	50
findMany orderBy	2.04ms±0.073	0.581ms±0.022	0.706ms	3.50x1.22x D	***	13.5%	50
findMany pagination	0.156ms±0.0097	0.138ms±0.110	0.232ms	~1.00x	≈	201.6%	50
findMany select	0.113ms±0.0050	0.120ms±0.0053	0.101ms	~1.00x	≈	15.6%	50
findMany relation some	0.527ms±0.0044	0.404ms±0.0050	—	~1.00x	≈	4.4%	50
findMany relation every	0.515ms±0.0058	0.360ms±0.0069	—	~1.00x	≈	7.0%	50
findMany relation none	9.89ms±0.305	8.53ms±0.378	—	1.16x	***	16.7%	50
findMany nested relation	0.578ms±0.016	0.392ms±0.045	—	~1.00x	≈	38.6%	50
findMany complex	0.799ms±0.016	0.461ms±0.083	0.587ms	~1.00x	≈	57.8%	50
findFirst	0.132ms±0.0047	0.101ms±0.011	0.197ms	~1.00x	≈	35.7%	50
findFirst skip	0.135ms±0.048	0.105ms±0.025	0.189ms	~1.00x	≈	64.2%	50
findUnique id	0.107ms±0.024	0.070ms±0.015	0.130ms	~1.00x	≈	65.1%	50
findUnique email	0.106ms±0.0050	0.065ms±0.0097	0.128ms	~1.00x	≈	47.4%	50
count	0.079ms±0.0036	0.060ms±0.0028	0.059ms	~1.00x	≈	16.4%	50
count where	0.398ms±0.0086	0.239ms±0.013	0.272ms	~1.00x	≈	18.4%	50
aggregate count	0.174ms±0.0019	0.146ms±0.0022	—	~1.00x	≈	5.1%	50
aggregate sum/avg	0.265ms±0.0033	0.248ms±0.0033	—	~1.00x	≈	5.0%	50
aggregate where	0.369ms±0.0047	0.229ms±0.0030	—	~1.00x	≈	4.7%	50
aggregate min/max	0.256ms±0.0036	0.247ms±0.0030	—	~1.00x	≈	4.6%	50
aggregate complete	0.325ms±0.0036	0.284ms±0.0033	—	~1.00x	≈	4.3%	50
groupBy	0.360ms±0.022	0.349ms±0.0072	—	~1.00x	≈	7.4%	50
groupBy count	0.368ms±0.0072	0.359ms±0.012	—	~1.00x	≈	11.7%	50
groupBy multi	0.460ms±0.0089	0.453ms±0.016	—	~1.00x	≈	12.3%	50
groupBy having	0.432ms±0.0075	0.409ms±0.050	—	~1.00x	≈	41.0%	50
groupBy + where	0.425ms±0.0030	0.278ms±0.0039	—	~1.00x	≈	5.0%	50
_count via include	0.606ms±0.0055	0.533ms±0.012	—	~1.00x	≈	7.8%	50
_count via include + relations	1.16ms±0.011	0.635ms±0.0091	—	~1.00x	≈	5.1%	50
groupBy aggregates	0.479ms±0.0064	0.465ms±0.0044	—	~1.00x	≈	3.4%	50
groupBy min/max	0.462ms±0.051	0.452ms±0.0044	—	~1.00x	≈	3.6%	50
include list + nested to-one	0.941ms±0.015	0.324ms±0.013	—	~1.00x	≈	13.8%	50
include list + nullable to-one	1.29ms±0.196	1.53ms±0.027	—	~1.00x	≈	6.2%	50
include posts	1.36ms±0.143	1.29ms±0.111	2.55ms	~1.00x	≈	29.7%	50
include profile	0.316ms±0.027	0.248ms±0.029	0.462ms	~1.00x	≈	35.6%	50
include 3 levels	1.80ms±0.210	1.92ms±0.190	2.50ms	~1.00x	≈	32.2%	50
include 4 levels	5.99ms±0.553	4.24ms±0.269	5.54ms	1.41x1.31x D	***	22.9%	50
include + where	2.43ms±0.202	2.31ms±0.251	1.58ms	~1.00x	≈	35.6%	50
include + select nested	1.90ms±0.238	2.15ms±0.330	4.86ms	~1.00x	≈	52.6%	50
findMany startsWith	0.124ms±0.011	0.089ms±0.0075	0.183ms	~1.00x	≈	27.5%	50
findMany endsWith	0.235ms±0.011	0.165ms±0.0036	0.252ms	~1.00x	≈	8.0%	50
findMany NOT contains	0.218ms±0.0069	0.158ms±0.0047	0.234ms	~1.00x	≈	10.7%	50
findMany LIKE	0.081ms±0.0039	0.051ms±0.0022	0.103ms	~1.00x	≈	14.9%	50
findMany <	10.36ms±0.265	6.60ms±0.759	8.31ms	1.57x1.26x D	***	37.6%	50
findMany <=	11.00ms±0.608	6.71ms±0.552	8.80ms	1.64x1.31x D	***	28.2%	50
findMany >	7.08ms±0.345	4.40ms±0.411	6.08ms	1.61x1.38x D	***	30.2%	50
findMany NOT IN	0.381ms±0.064	0.241ms±0.042	0.437ms	~1.00x	≈	55.8%	50
findMany isNot null	0.212ms±0.011	0.168ms±0.011	0.197ms	~1.00x	≈	22.0%	50
orderBy multi-field	2.04ms±0.039	0.560ms±0.016	0.662ms	3.64x1.18x D	***	10.1%	50
orderBy relation to-one	1.87ms±0.049	1.39ms±0.052	—	~1.00x	≈	13.0%	50
orderBy relation + scalar	1.95ms±0.057	1.46ms±0.057	—	~1.00x	≈	13.4%	50
orderBy nullable relation	1.59ms±0.032	1.15ms±0.018	—	~1.00x	≈	5.8%	50
orderBy deep relation	2.09ms±0.121	1.61ms±0.072	—	~1.00x	≈	15.3%	50
orderBy deep relation + scalar	2.61ms±0.124	2.10ms±0.305	—	~1.00x	≈	45.9%	50
distinct status	7.73ms±0.274	1.28ms±0.051	—	6.03x	***	13.9%	50
distinct multi	11.00ms±0.363	2.54ms±0.220	—	4.33x	***	28.5%	50
cursor composite tuple	1.39ms±0.0067	1.09ms±0.0078	—	~1.00x	≈	2.5%	50
cursor composite desc	1.84ms±0.114	1.45ms±0.080	—	~1.00x	≈	18.9%	50
cursor prefix of orderBy	2.61ms±0.123	2.01ms±0.099	—	~1.00x	≈	17.7%	50
select + include	1.06ms±0.039	0.335ms±0.022	0.594ms	~1.00x	≈	22.8%	50
_count relation	1.66ms±0.123	1.80ms±0.107	—	~1.00x	≈	22.1%	50
_count multi-relation	0.737ms±0.041	0.715ms±0.041	—	~1.00x	≈	20.0%	50
_count inside include	4.12ms±0.256	2.85ms±0.151	—	1.45x	***	19.5%	50
_count inside nested select	1.54ms±0.078	1.45ms±0.091	—	~1.00x	≈	21.3%	50
_count deep include	2.19ms±0.135	1.76ms±0.083	—	~1.00x	≈	16.0%	50
ILIKE special chars	0.184ms±0.028	0.139ms±0.025	—	~1.00x	≈	52.5%	50
LIKE case sensitive	0.135ms±0.0061	0.099ms±0.0044	—	~1.00x	≈	15.2%	50
findMany Date range	0.150ms±0.0067	0.113ms±0.0055	0.205ms	~1.00x	≈	16.7%	50
count Date range	0.353ms±0.0050	0.316ms±0.0047	0.343ms	~1.00x	≈	5.4%	50
findMany Date gte	0.145ms±0.0078	0.105ms±0.0067	0.177ms	~1.00x	≈	21.7%	50
depth-1 low-fan	0.211ms±0.049	0.155ms±0.029	—	~1.00x	≈	55.6%	50
depth-1 mid-fan	1.03ms±0.123	0.959ms±0.142	—	~1.00x	≈	47.4%	50
depth-1 high-fan	0.840ms±0.107	0.791ms±0.011	—	~1.00x	≈	4.7%	50
depth-1 wide	0.789ms±0.120	0.826ms±0.037	—	~1.00x	≈	15.5%	50
depth-1 unbound	17.14ms±0.884	8.82ms±0.673	—	1.94x	***	25.0%	50
depth-2	3.76ms±0.234	3.72ms±0.175	—	~1.00x	≈	17.0%	50
depth-2 paginated Project→tasks	2.36ms±0.296	2.73ms±0.275	—	~1.00x	≈	40.8%	50
depth-2 high-fan	3.13ms±0.208	3.08ms±0.108	—	~1.00x	≈	12.4%	50
depth-2 wide	3.11ms±0.244	3.16ms±0.207	—	~1.00x	≈	22.5%	50
depth-2 unbound	30.14ms±1.02	17.11ms±0.944	—	1.76x	***	19.2%	50
depth-3 unbound	10.80ms±0.443	8.68ms±0.371	—	1.25x	***	15.1%	50
depth-3 paginated	1.19ms±0.112	1.10ms±0.105	—	~1.00x	≈	31.3%	50
depth-4 unbound	4.88ms±0.333	4.59ms±0.360	—	~1.00x	≈	28.9%	50
depth-4 paginated	1.31ms±0.047	0.976ms±0.068	—	~1.00x	≈	23.6%	50
findFirst depth-2	0.853ms±0.024	0.831ms±0.078	—	~1.00x	≈	31.1%	50
findUnique depth-2 Project→tasks→comment	0.846ms±0.093	0.727ms±0.021	—	~1.00x	≈	9.8%	50
complex nested select	2.88ms±0.093	1.82ms±0.035	—	1.58x	***	6.9%	50
ultra deep query	11.34ms±0.241	7.04ms±0.089	—	1.61x	***	4.5%	50
transaction: (3 operations)	3.29ms±0.313	1.56ms±0.311	—	2.12x	***	62.9%	50

*** p < 0.001** p < 0.01* p < 0.05ns not significant≈ within 1ms noise floor± 95% CICV% > 15% = high variance

PostgreSQLPrisma v8 (8.1.0-dev.1)

Avg 1.54x vs PrismaAvg 1.63x vs Drizzle17/95 significant wins78 within noise floor

Sep 30, 2026

Test	Prisma	prisma-sql	Drizzle	Speedup	Sig.	CV%	n
findMany basic	0.279ms±0.016	0.236ms±0.024	0.275ms	~1.00x	≈	32.6%	50
findMany where =	0.362ms±0.186	0.245ms±0.075	0.413ms	~1.00x	≈	89.0%	50
findMany where >=	8.32ms±0.363	4.81ms±0.542	6.99ms	1.73x1.45x D	***	37.4%	50
findMany where IN	0.383ms±0.014	0.254ms±0.011	0.420ms	~1.00x	≈	14.9%	50
findMany where null	0.166ms±0.013	0.137ms±0.084	0.150ms	~1.00x	≈	166.4%	50
findMany ILIKE	0.115ms±0.0053	0.083ms±0.0033	0.146ms	~1.00x	≈	14.0%	50
findMany AND	0.973ms±0.077	0.550ms±0.043	0.713ms	~1.00x	≈	26.8%	50
findMany OR	7.35ms±0.505	4.16ms±0.242	5.12ms	1.77x1.23x D	***	20.6%	50
findMany NOT	0.505ms±0.102	0.273ms±0.017	0.521ms	~1.00x	≈	21.2%	50
findMany orderBy	3.11ms±0.101	1.12ms±0.072	1.32ms	2.78x1.18x D	***	24.9%	50
findMany pagination	0.167ms±0.0100	0.135ms±0.013	0.246ms	~1.00x	≈	31.6%	50
findMany select	0.117ms±0.138	0.119ms±0.0083	0.101ms	~1.00x	≈	23.5%	50
findMany relation some	0.543ms±0.025	0.389ms±0.019	—	~1.00x	≈	16.4%	50
findMany relation every	0.551ms±0.0078	0.388ms±0.0069	—	~1.00x	≈	6.3%	50
findMany relation none	10.52ms±0.391	8.40ms±0.373	—	1.25x	***	16.8%	50
findMany nested relation	0.558ms±0.011	0.371ms±0.0069	—	~1.00x	≈	6.7%	50
findMany complex	0.769ms±0.020	0.451ms±0.0089	0.584ms	~1.00x	≈	7.1%	50
findFirst	0.135ms±0.0072	0.101ms±0.0039	0.195ms	~1.00x	≈	13.3%	50
findFirst skip	0.110ms±0.0072	0.077ms±0.0033	0.151ms	~1.00x	≈	15.7%	50
findUnique id	0.103ms±0.0028	0.065ms±0.0014	0.127ms	~1.00x	≈	8.2%	50
findUnique email	0.099ms±0.0025	0.066ms±0.0028	0.125ms	~1.00x	≈	14.9%	50
count	0.081ms±0.0039	0.062ms±0.0019	0.061ms	~1.00x	≈	11.1%	50
count where	0.381ms±0.0086	0.241ms±0.0036	0.273ms	~1.00x	≈	5.5%	50
aggregate count	0.178ms±0.0033	0.148ms±0.0025	—	~1.00x	≈	5.9%	50
aggregate sum/avg	0.268ms±0.0050	0.250ms±0.0033	—	~1.00x	≈	4.6%	50
aggregate where	0.369ms±0.0039	0.237ms±0.0033	—	~1.00x	≈	5.2%	50
aggregate min/max	0.263ms±0.0039	0.252ms±0.0036	—	~1.00x	≈	5.1%	50
aggregate complete	0.331ms±0.011	0.294ms±0.065	—	~1.00x	≈	70.0%	50
groupBy	0.346ms±0.0061	0.337ms±0.0047	—	~1.00x	≈	5.1%	50
groupBy count	0.367ms±0.0047	0.355ms±0.0047	—	~1.00x	≈	4.8%	50
groupBy multi	0.463ms±0.0091	0.442ms±0.010	—	~1.00x	≈	8.1%	50
groupBy having	0.407ms±0.017	0.379ms±0.013	—	~1.00x	≈	11.7%	50
groupBy + where	0.393ms±0.016	0.257ms±0.0064	—	~1.00x	≈	8.6%	50
_count via include	0.564ms±0.013	0.484ms±0.0097	—	~1.00x	≈	7.1%	50
_count via include + relations	1.19ms±0.089	0.657ms±0.016	—	~1.00x	≈	8.8%	50
groupBy aggregates	0.462ms±0.013	0.455ms±0.0075	—	~1.00x	≈	5.9%	50
groupBy min/max	0.454ms±0.0050	0.446ms±0.0058	—	~1.00x	≈	4.8%	50
include list + nested to-one	0.967ms±0.060	0.339ms±0.014	—	~1.00x	≈	14.2%	50
include list + nullable to-one	1.33ms±0.110	1.64ms±0.096	—	~1.00x	≈	20.4%	50
include posts	1.24ms±0.160	1.21ms±0.116	2.31ms	~1.00x	≈	30.5%	50
include profile	0.373ms±0.031	0.287ms±0.039	0.519ms	~1.00x	≈	38.7%	50
include 3 levels	2.78ms±0.435	2.80ms±0.203	3.36ms	~1.00x	≈	29.4%	50
include 4 levels	1.64ms±0.609	1.36ms±0.245	1.83ms	~1.00x	≈	47.0%	50
include + where	1.84ms±0.089	1.76ms±0.172	4.45ms	~1.00x	≈	34.6%	50
include + select nested	0.957ms±0.165	1.02ms±0.161	2.90ms	~1.00x	≈	43.0%	50
findMany startsWith	0.127ms±0.011	0.097ms±0.0061	0.194ms	~1.00x	≈	21.8%	50
findMany endsWith	0.260ms±0.0094	0.189ms±0.080	0.295ms	~1.00x	≈	119.6%	50
findMany NOT contains	0.225ms±0.0058	0.165ms±0.0030	0.242ms	~1.00x	≈	6.3%	50
findMany LIKE	0.090ms±0.0061	0.061ms±0.0042	0.121ms	~1.00x	≈	24.6%	50
findMany <	11.95ms±1.04	6.55ms±0.246	8.71ms	1.82x1.33x D	***	13.2%	50
findMany <=	12.08ms±1.58	6.74ms±0.432	8.73ms	1.79x1.30x D	***	21.8%	50
findMany >	7.81ms±0.481	4.60ms±0.239	5.86ms	1.70x1.27x D	***	18.7%	50
findMany NOT IN	0.358ms±0.077	0.226ms±0.013	0.391ms	~1.00x	≈	19.2%	50
findMany isNot null	0.283ms±0.012	0.229ms±0.0094	0.253ms	~1.00x	≈	15.5%	50
orderBy multi-field	1.88ms±0.042	0.522ms±0.071	0.590ms	3.60x1.13x D	***	46.1%	50
orderBy relation to-one	1.69ms±0.021	1.29ms±0.031	—	~1.00x	≈	8.5%	50
orderBy relation + scalar	2.69ms±0.218	2.05ms±0.176	—	~1.00x	≈	30.7%	50
orderBy nullable relation	1.59ms±0.198	1.15ms±0.027	—	~1.00x	≈	8.3%	50
orderBy deep relation	1.91ms±0.018	1.50ms±0.013	—	~1.00x	≈	3.1%	50
orderBy deep relation + scalar	2.33ms±0.124	1.90ms±0.085	—	~1.00x	≈	14.8%	50
distinct status	7.76ms±0.096	1.30ms±0.091	—	5.98x	***	24.1%	50
distinct multi	11.01ms±0.071	2.47ms±0.035	—	4.45x	***	5.1%	50
cursor composite tuple	1.45ms±0.016	1.13ms±0.018	—	~1.00x	≈	5.7%	50
cursor composite desc	1.59ms±0.0100	1.27ms±0.0100	—	~1.00x	≈	2.8%	50
cursor prefix of orderBy	1.53ms±0.251	1.17ms±0.167	—	~1.00x	≈	40.3%	50
select + include	0.619ms±0.015	0.206ms±0.013	0.370ms	~1.00x	≈	21.4%	50
_count relation	0.581ms±0.0050	0.596ms±0.0055	—	~1.00x	≈	3.4%	50
_count multi-relation	0.193ms±0.0069	0.183ms±0.0044	—	~1.00x	≈	8.4%	50
_count inside include	0.910ms±0.020	0.633ms±0.013	—	~1.00x	≈	7.3%	50
_count inside nested select	1.37ms±0.020	1.18ms±0.017	—	~1.00x	≈	5.1%	50
_count deep include	1.52ms±0.015	1.20ms±0.020	—	~1.00x	≈	5.9%	50
ILIKE special chars	0.148ms±0.0053	0.120ms±0.0042	—	~1.00x	≈	11.8%	50
LIKE case sensitive	0.117ms±0.0028	0.087ms±0.0019	—	~1.00x	≈	7.4%	50
findMany Date range	1.28ms±0.012	0.139ms±0.032	0.549ms	9.27x3.96x D	***	62.7%	50
count Date range	0.389ms±0.0064	0.349ms±0.0067	0.409ms	~1.00x	≈	6.9%	50
findMany Date gte	0.169ms±0.0039	0.130ms±0.0028	0.227ms	~1.00x	≈	7.8%	50
depth-1 low-fan	0.199ms±0.0050	0.133ms±0.0036	—	~1.00x	≈	9.8%	50
depth-1 mid-fan	0.997ms±0.103	0.915ms±0.073	—	~1.00x	≈	27.5%	50
depth-1 high-fan	0.904ms±0.102	0.890ms±0.109	—	~1.00x	≈	41.3%	50
depth-1 wide	0.857ms±0.091	0.925ms±0.019	—	~1.00x	≈	7.4%	50
depth-1 unbound	17.36ms±0.533	9.64ms±0.963	—	1.80x	***	31.6%	50
depth-2	2.64ms±0.122	2.78ms±0.226	—	~1.00x	≈	28.5%	50
depth-2 paginated Project→tasks	0.991ms±0.075	1.17ms±0.112	—	~1.00x	≈	31.1%	50
depth-2 high-fan	1.31ms±0.145	1.32ms±0.096	—	~1.00x	≈	22.5%	50
depth-2 wide	2.47ms±0.216	2.45ms±0.193	—	~1.00x	≈	25.7%	50
depth-2 unbound	28.05ms±0.484	15.71ms±0.461	—	1.79x	***	10.5%	50
depth-3 unbound	8.51ms±0.281	7.07ms±0.491	—	1.20x	***	24.1%	50
depth-3 paginated	1.68ms±0.183	1.64ms±0.160	—	~1.00x	≈	31.7%	50
depth-4 unbound	6.55ms±0.301	6.09ms±0.342	—	~1.00x	≈	19.7%	50
depth-4 paginated	1.43ms±0.162	1.18ms±0.192	—	~1.00x	≈	47.1%	50
findFirst depth-2	0.987ms±0.094	0.971ms±0.019	—	~1.00x	≈	7.1%	50
findUnique depth-2 Project→tasks→comment	1.53ms±0.135	1.39ms±0.062	—	~1.00x	≈	15.8%	50
complex nested select	3.37ms±0.296	2.05ms±0.185	—	1.64x	***	27.8%	50
ultra deep query	12.44ms±0.689	7.71ms±0.293	—	1.61x	***	13.2%	50
transaction: (3 operations)	10.90ms±0.134	9.07ms±0.081	—	1.20x	***	3.2%	50

*** p < 0.001** p < 0.01* p < 0.05ns not significant≈ within 1ms noise floor± 95% CICV% > 15% = high variance

SQLitePrisma v6 (6.19.3)

Avg 4.80x vs PrismaAvg 1.79x vs Drizzle22/84 significant wins62 within noise floor

Sep 30, 2026

Test	Prisma	prisma-sql	Drizzle	Speedup	Sig.	CV%	n
findMany basic	0.449ms±0.019	0.070ms±0.0083	0.116ms	~1.00x	≈	37.1%	50
findMany where =	0.380ms±0.0075	0.055ms±0.0025	0.093ms	~1.00x	≈	15.4%	50
findMany where >=	13.30ms±0.272	1.33ms±0.039	2.14ms	10.03x1.61x D	***	10.3%	50
findMany where IN	0.432ms±0.012	0.063ms±0.0033	0.114ms	~1.00x	≈	18.2%	50
findMany where null	0.167ms±0.0047	0.025ms±0.0030	0.057ms	~1.00x	≈	37.9%	50
findMany AND	1.55ms±0.141	0.302ms±0.041	0.424ms	5.12x1.40x D	***	42.4%	50
findMany OR	10.21ms±0.128	1.12ms±0.023	1.80ms	9.12x1.61x D	***	7.5%	50
findMany NOT	0.436ms±0.017	0.055ms±0.0033	0.103ms	~1.00x	≈	20.5%	50
findMany orderBy	2.37ms±0.033	1.83ms±0.021	1.88ms	~1.00x	≈	4.1%	50
findMany pagination	0.147ms±0.012	0.037ms±0.0028	0.066ms	~1.00x	≈	24.7%	50
findMany select	0.173ms±0.0044	0.047ms±0.0019	0.041ms	~1.00x	≈	15.2%	50
findMany relation some	4.56ms±0.020	0.446ms±0.0094	—	10.21x	***	7.5%	50
findMany relation every	9.74ms±0.108	6.35ms±0.059	—	1.53x	***	3.3%	50
findMany relation none	169.69ms±3.84	2.81ms±0.108	—	60.47x	***	6.0%	10
findMany nested relation	1.04ms±0.018	0.316ms±0.022	—	~1.00x	≈	23.5%	50
findMany complex	0.814ms±0.084	0.405ms±0.0086	0.454ms	~1.00x	≈	7.4%	50
findFirst	0.163ms±0.021	0.035ms±0.0072	0.074ms	~1.00x	≈	59.4%	50
findFirst skip	0.184ms±0.126	0.036ms±0.0083	0.081ms	~1.00x	≈	66.4%	50
findUnique id	0.134ms±0.0033	0.032ms±0.0017	0.059ms	~1.00x	≈	17.1%	50
findUnique email	0.123ms±0.0017	0.029ms±0.0008	0.056ms	~1.00x	≈	10.5%	50
count	0.068ms±0.0019	0.016ms±0.0008	0.019ms	~1.00x	≈	18.3%	50
count where	0.268ms±0.0025	0.164ms±0.0019	0.169ms	~1.00x	≈	4.2%	50
_count via include	0.537ms±0.0067	0.232ms±0.0061	—	~1.00x	≈	9.4%	50
_count via include + relations	1.06ms±0.0055	0.276ms±0.0055	—	~1.00x	≈	7.2%	50
aggregate count	0.144ms±0.0044	0.027ms±0.0017	—	~1.00x	≈	20.4%	50
aggregate sum/avg	0.274ms±0.0022	0.168ms±0.0019	—	~1.00x	≈	4.1%	50
aggregate where	0.262ms±0.0025	0.166ms±0.0030	—	~1.00x	≈	6.5%	50
aggregate min/max	0.306ms±0.0022	0.175ms±0.0025	—	~1.00x	≈	5.1%	50
aggregate complete	0.390ms±0.0025	0.234ms±0.0028	—	~1.00x	≈	4.4%	50
groupBy	0.572ms±0.0089	0.454ms±0.0061	—	~1.00x	≈	4.8%	50
groupBy count	0.567ms±0.0061	0.453ms±0.0047	—	~1.00x	≈	3.8%	50
groupBy multi	1.17ms±0.0072	0.971ms±0.0075	—	~1.00x	≈	2.8%	50
groupBy having	0.635ms±0.0091	0.485ms±0.0047	—	~1.00x	≈	3.5%	50
groupBy + where	0.325ms±0.065	0.204ms±0.0036	—	~1.00x	≈	6.2%	50
groupBy aggregates	0.703ms±0.0078	0.578ms±0.0044	—	~1.00x	≈	2.8%	50
groupBy min/max	0.718ms±0.0055	0.583ms±0.0036	—	~1.00x	≈	2.2%	50
include posts	2.00ms±0.041	0.421ms±0.021	1.34ms	4.75x3.18x D	***	16.9%	50
include profile	0.333ms±0.0097	0.066ms±0.0022	0.191ms	~1.00x	≈	11.9%	50
include 3 levels	1.11ms±0.032	0.642ms±0.0064	0.817ms	~1.00x	≈	3.6%	50
include 4 levels	1.14ms±0.0075	0.423ms±0.0050	0.634ms	~1.00x	≈	4.1%	50
include + where	0.777ms±0.021	0.289ms±0.0058	0.271ms	~1.00x	≈	7.1%	50
include + select nested	0.778ms±0.026	0.317ms±0.0067	0.953ms	~1.00x	≈	7.5%	50
findMany startsWith	0.157ms±0.0064	0.035ms±0.0014	0.059ms	~1.00x	≈	14.2%	50
findMany endsWith	0.459ms±0.0064	0.067ms±0.0022	0.106ms	~1.00x	≈	11.1%	50
findMany NOT contains	0.455ms±0.058	0.054ms±0.0019	0.095ms	~1.00x	≈	11.6%	50
findMany LIKE	0.156ms±0.0022	0.034ms±0.0017	0.057ms	~1.00x	≈	17.2%	50
findMany <	21.45ms±0.104	2.21ms±0.316	3.56ms	9.69x1.61x D	***	41.4%	50
findMany <=	22.25ms±0.096	2.23ms±0.235	3.69ms	9.96x1.65x D	***	33.3%	50
findMany >	11.74ms±0.116	1.15ms±0.161	1.92ms	10.21x1.67x D	***	45.5%	50
findMany NOT IN	0.444ms±0.014	0.058ms±0.011	0.115ms	~1.00x	≈	56.6%	50
findMany isNot null	0.426ms±0.0055	0.051ms±0.0017	0.093ms	~1.00x	≈	11.6%	50
orderBy multi-field	0.671ms±0.069	0.350ms±0.0064	0.402ms	~1.00x	≈	6.5%	50
orderBy relation to-one	0.641ms±0.0055	0.319ms±0.0086	—	~1.00x	≈	9.4%	50
orderBy relation + scalar	0.714ms±0.0055	0.391ms±0.0039	—	~1.00x	≈	3.7%	50
orderBy nullable relation	0.608ms±0.0069	0.321ms±0.0025	—	~1.00x	≈	2.7%	50
orderBy deep relation	0.710ms±0.0086	0.380ms±0.0044	—	~1.00x	≈	4.3%	50
orderBy deep relation + scalar	1.10ms±0.010	0.694ms±0.0078	—	~1.00x	≈	4.0%	50
distinct status	10.99ms±0.072	4.09ms±0.035	—	2.68x	***	3.0%	50
distinct multi	12.12ms±0.092	4.96ms±0.013	—	2.44x	***	1.0%	50
select + include	0.537ms±0.0061	0.066ms±0.0028	0.133ms	~1.00x	≈	14.1%	50
_count relation	0.611ms±0.0061	0.353ms±0.0039	—	~1.00x	≈	3.9%	50
_count multi-relation	0.132ms±0.0061	0.045ms±0.0036	—	~1.00x	≈	25.1%	50
_count inside include	0.861ms±0.0078	0.341ms±0.0058	—	~1.00x	≈	6.1%	50
_count inside nested select	1.54ms±0.010	0.587ms±0.0072	—	~1.00x	≈	4.5%	50
_count deep include	1.33ms±0.0083	0.730ms±0.0089	—	~1.00x	≈	4.3%	50
findMany Date range	0.259ms±0.0042	0.051ms±0.0014	0.230ms	~1.00x	≈	9.3%	50
count Date range	0.303ms±0.0050	0.180ms±0.0025	0.182ms	~1.00x	≈	5.1%	50
findMany Date gte	0.242ms±0.0033	0.049ms±0.0019	0.102ms	~1.00x	≈	13.1%	50
depth-1 low-fan Project→labels	0.257ms±0.0067	0.069ms±0.0042	—	~1.00x	≈	19.7%	50
depth-1 mid-fan Project→tasks	1.91ms±0.022	0.367ms±0.0083	—	5.21x	***	8.0%	50
depth-1 high-fan User→assignedTasks	1.65ms±0.062	0.332ms±0.0086	—	4.96x	***	9.2%	50
depth-1 wide	1.43ms±0.053	0.337ms±0.0061	—	4.25x	***	6.6%	50
depth-1 no-limit Project→tasks	33.39ms±0.407	3.56ms±0.378	—	9.39x	***	34.4%	50
depth-2	1.77ms±0.045	0.471ms±0.012	—	3.76x	***	9.4%	50
depth-2 paginated	0.896ms±0.0086	0.663ms±0.0067	—	~1.00x	≈	3.6%	50
depth-2 high-fan	1.56ms±0.034	0.416ms±0.012	—	3.76x	***	9.8%	50
depth-2 wide+attach+activity	1.86ms±0.031	0.532ms±0.013	—	3.49x	***	8.4%	50
depth-2 no-limit	49.34ms±0.687	6.75ms±0.300	—	7.31x	***	7.2%	10
depth-3	9.70ms±0.121	1.55ms±0.032	—	6.25x	***	7.3%	50
depth-3 paginated	1.07ms±0.0094	0.604ms±0.0067	—	~1.00x	≈	4.0%	50
depth-4	5.66ms±0.217	1.04ms±0.038	—	5.44x	***	12.8%	50
depth-4 paginated	1.12ms±0.079	0.418ms±0.0067	—	~1.00x	≈	5.6%	50
findFirst depth-2	0.906ms±0.0080	0.285ms±0.0083	—	~1.00x	≈	10.2%	50
findUnique depth-2	0.927ms±0.051	0.290ms±0.0083	—	~1.00x	≈	9.9%	50

*** p < 0.001** p < 0.01* p < 0.05ns not significant≈ within 1ms noise floor± 95% CICV% > 15% = high variance

SQLitePrisma v7 (7.10.0)

Avg 2.29x vs PrismaAvg 1.84x vs Drizzle13/84 significant wins71 within noise floor

Sep 30, 2026

Test	Prisma	prisma-sql	Drizzle	Speedup	Sig.	CV%	n
findMany basic	0.147ms±0.0033	0.056ms±0.0030	0.105ms	~1.00x	≈	17.1%	50
findMany where =	0.123ms±0.0025	0.051ms±0.0011	0.090ms	~1.00x	≈	6.8%	50
findMany where >=	3.70ms±0.269	1.21ms±0.017	2.15ms	3.05x1.77x D	***	5.0%	50
findMany where IN	0.140ms±0.251	0.057ms±0.0028	0.101ms	~1.00x	≈	17.1%	50
findMany where null	0.061ms±0.0019	0.024ms±0.0008	0.050ms	~1.00x	≈	12.3%	50
findMany AND	0.566ms±0.169	0.297ms±0.0055	0.410ms	~1.00x	≈	6.8%	50
findMany OR	3.08ms±0.218	1.09ms±0.186	1.86ms	2.82x1.70x D	***	53.4%	50
findMany NOT	0.147ms±0.239	0.058ms±0.0036	0.123ms	~1.00x	≈	21.1%	50
findMany orderBy	1.72ms±0.011	1.67ms±0.0097	1.71ms	~1.00x	≈	2.1%	50
findMany pagination	0.065ms±0.0025	0.037ms±0.012	0.060ms	~1.00x	≈	93.4%	50
findMany select	0.059ms±0.0011	0.046ms±0.0014	0.039ms	~1.00x	≈	10.4%	50
findMany relation some	0.383ms±0.012	0.430ms±0.100	—	~1.00x	≈	74.8%	50
findMany relation every	9.48ms±0.212	6.42ms±0.085	—	1.48x	***	4.7%	50
findMany relation none	126.50ms±8.87	3.10ms±0.084	—	40.78x	***	4.3%	10
findMany nested relation	0.369ms±0.012	0.304ms±0.0064	—	~1.00x	≈	7.3%	50
findMany complex	0.549ms±0.156	0.509ms±0.103	0.530ms	~1.00x	≈	55.5%	50
findFirst	0.065ms±0.0069	0.027ms±0.0014	0.060ms	~1.00x	≈	18.7%	50
findFirst skip	0.075ms±0.0080	0.029ms±0.0042	0.076ms	~1.00x	≈	43.6%	50
findUnique id	0.063ms±0.0025	0.023ms±0.0006	0.051ms	~1.00x	≈	10.0%	50
findUnique email	0.062ms±0.0030	0.024ms±0.0008	0.052ms	~1.00x	≈	13.8%	50
count	0.033ms±0.0014	0.014ms±0.0003	0.014ms	~1.00x	≈	10.2%	50
count where	0.201ms±0.0058	0.166ms±0.0025	0.168ms	~1.00x	≈	5.1%	50
_count via include	0.416ms±0.0089	0.224ms±0.0086	—	~1.00x	≈	13.5%	50
_count via include + relations	0.759ms±0.016	0.269ms±0.011	—	~1.00x	≈	14.1%	50
aggregate count	0.074ms±0.146	0.024ms±0.0022	—	~1.00x	≈	28.5%	50
aggregate sum/avg	0.259ms±0.142	0.195ms±0.134	—	~1.00x	≈	120.6%	50
aggregate where	0.201ms±0.073	0.174ms±0.057	—	~1.00x	≈	86.9%	50
aggregate min/max	0.206ms±0.0039	0.172ms±0.0030	—	~1.00x	≈	6.3%	50
aggregate complete	0.274ms±0.0097	0.238ms±0.0033	—	~1.00x	≈	5.1%	50
groupBy	0.475ms±0.0058	0.447ms±0.0069	—	~1.00x	≈	5.5%	50
groupBy count	0.494ms±0.016	0.467ms±0.0097	—	~1.00x	≈	7.3%	50
groupBy multi	0.987ms±0.020	0.973ms±0.019	—	~1.00x	≈	6.9%	50
groupBy having	0.503ms±0.0047	0.478ms±0.0044	—	~1.00x	≈	3.3%	50
groupBy + where	0.219ms±0.0014	0.198ms±0.0011	—	~1.00x	≈	2.2%	50
groupBy aggregates	0.577ms±0.0055	0.552ms±0.0033	—	~1.00x	≈	2.2%	50
groupBy min/max	0.584ms±0.0067	0.561ms±0.0083	—	~1.00x	≈	5.3%	50
include posts	0.834ms±0.109	0.435ms±0.029	1.35ms	~1.00x	≈	22.4%	50
include profile	0.138ms±0.128	0.065ms±0.0044	0.194ms	~1.00x	≈	22.1%	50
include 3 levels	0.609ms±0.148	0.566ms±0.014	0.762ms	~1.00x	≈	8.7%	50
include 4 levels	0.682ms±0.104	0.424ms±0.027	0.641ms	~1.00x	≈	21.2%	50
include + where	0.399ms±0.0053	0.271ms±0.0078	0.264ms	~1.00x	≈	9.9%	50
include + select nested	0.344ms±0.0072	0.302ms±0.085	0.971ms	~1.00x	≈	87.0%	50
findMany startsWith	0.056ms±0.0011	0.033ms±0.0003	0.055ms	~1.00x	≈	3.9%	50
findMany endsWith	0.143ms±0.0036	0.069ms±0.0047	0.118ms	~1.00x	≈	23.5%	50
findMany NOT contains	0.142ms±0.0014	0.057ms±0.0008	0.099ms	~1.00x	≈	4.8%	50
findMany LIKE	0.055ms±0.0014	0.032ms±0.0003	0.052ms	~1.00x	≈	4.4%	50
findMany <	6.81ms±4.28	2.47ms±0.428	3.96ms	2.75x1.60x D	**	52.3%	50
findMany <=	6.86ms±0.959	2.48ms±0.433	4.11ms	2.76x1.66x D	***	48.9%	50
findMany >	3.85ms±0.589	1.35ms±0.229	2.19ms	2.85x1.62x D	***	50.0%	50
findMany NOT IN	0.142ms±0.0036	0.056ms±0.0061	0.103ms	~1.00x	≈	35.5%	50
findMany isNot null	0.135ms±0.0028	0.054ms±0.0030	0.096ms	~1.00x	≈	20.0%	50
orderBy multi-field	0.437ms±0.0075	0.366ms±0.0067	0.421ms	~1.00x	≈	6.5%	50
orderBy relation to-one	0.392ms±0.0050	0.327ms±0.0086	—	~1.00x	≈	9.3%	50
orderBy relation + scalar	0.460ms±0.0028	0.398ms±0.0033	—	~1.00x	≈	2.9%	50
orderBy nullable relation	0.394ms±0.015	0.335ms±0.016	—	~1.00x	≈	16.3%	50
orderBy deep relation	2.14ms±1.01	1.59ms±1.77	—	~1.00x	≈	178.6%	50
orderBy deep relation + scalar	0.809ms±0.030	0.721ms±0.054	—	~1.00x	≈	25.0%	50
distinct status	6.38ms±2.53	4.55ms±0.932	—	1.40x	**	58.1%	50
distinct multi	7.42ms±0.707	5.78ms±0.551	—	1.28x	***	30.9%	50
select + include	0.300ms±0.0094	0.063ms±0.0083	0.132ms	~1.00x	≈	41.9%	50
_count relation	0.445ms±0.059	0.396ms±0.053	—	~1.00x	≈	41.8%	50
_count multi-relation	0.091ms±0.0064	0.045ms±0.0006	—	~1.00x	≈	5.3%	50
_count inside include	0.747ms±0.045	0.381ms±0.040	—	~1.00x	≈	32.5%	50
_count inside nested select	1.43ms±0.058	0.694ms±0.068	—	~1.00x	≈	31.4%	50
_count deep include	1.12ms±0.046	0.841ms±0.122	—	~1.00x	≈	47.5%	50
findMany Date range	0.102ms±0.0055	0.054ms±0.0047	0.311ms	~1.00x	≈	29.2%	50
count Date range	0.445ms±0.066	0.371ms±0.050	0.286ms	~1.00x	≈	42.0%	50
findMany Date gte	0.099ms±0.012	0.052ms±0.0072	0.089ms	~1.00x	≈	43.8%	50
depth-1 low-fan Project→labels	0.120ms±0.014	0.067ms±0.0044	—	~1.00x	≈	22.6%	50
depth-1 mid-fan Project→tasks	0.875ms±0.184	0.449ms±0.457	—	~1.00x	≈	221.0%	50
depth-1 high-fan User→assignedTasks	0.785ms±0.188	0.424ms±0.041	—	~1.00x	≈	31.4%	50
depth-1 wide	0.676ms±0.210	0.380ms±0.051	—	~1.00x	≈	41.2%	50
depth-1 no-limit Project→tasks	13.27ms±1.11	4.11ms±0.355	—	3.23x	***	27.1%	50
depth-2	0.808ms±0.422	0.468ms±0.212	—	~1.00x	≈	124.7%	50
depth-2 paginated	0.552ms±0.038	0.653ms±0.020	—	~1.00x	≈	10.6%	50
depth-2 high-fan	0.724ms±0.032	0.426ms±0.128	—	~1.00x	≈	88.3%	50
depth-2 wide+attach+activity	1.01ms±0.147	0.544ms±0.131	—	~1.00x	≈	73.6%	50
depth-2 no-limit	18.70ms±0.737	7.54ms±0.729	—	2.48x	***	29.4%	50
depth-3	3.63ms±0.429	1.54ms±0.348	—	2.36x	***	66.1%	50
depth-3 paginated	1.27ms±0.412	0.943ms±0.222	—	~1.00x	≈	63.7%	50
depth-4	2.67ms±0.426	1.23ms±0.214	—	2.16x	***	53.9%	50
depth-4 paginated	0.669ms±0.135	0.420ms±0.027	—	~1.00x	≈	20.9%	50
findFirst depth-2	0.508ms±0.149	0.296ms±0.019	—	~1.00x	≈	20.4%	50
findUnique depth-2	0.496ms±0.096	0.287ms±0.015	—	~1.00x	≈	17.4%	50

*** p < 0.001** p < 0.01* p < 0.05ns not significant≈ within 1ms noise floor± 95% CICV% > 15% = high variance

SQLitePrisma v8 (8.1.0-dev.1)

Avg 2.10x vs PrismaAvg 1.63x vs Drizzle14/84 significant wins70 within noise floor

Sep 30, 2026

Test	Prisma	prisma-sql	Drizzle	Speedup	Sig.	CV%	n
findMany basic	0.450ms±0.927	0.151ms±0.048	0.268ms	~1.00x	≈	72.4%	50
findMany where =	0.244ms±0.070	0.114ms±0.040	0.170ms	~1.00x	≈	80.5%	50
findMany where >=	7.43ms±1.10	3.18ms±1.15	4.17ms	2.34x1.31x D	***	98.5%	50
findMany where IN	0.284ms±0.065	0.117ms±0.041	0.169ms	~1.00x	≈	85.0%	50
findMany where null	0.138ms±0.035	0.052ms±0.023	0.082ms	~1.00x	≈	90.5%	50
findMany AND	1.61ms±0.206	0.836ms±0.120	0.976ms	~1.00x	≈	45.8%	50
findMany OR	5.29ms±0.590	2.24ms±0.382	3.25ms	2.36x1.45x D	***	55.5%	50
findMany NOT	0.197ms±0.040	0.081ms±0.026	0.126ms	~1.00x	≈	76.2%	50
findMany orderBy	4.11ms±0.759	3.79ms±0.449	3.61ms	~1.00x	≈	40.2%	50
findMany pagination	0.539ms±0.297	0.394ms±0.171	0.369ms	~1.00x	≈	103.6%	50
findMany select	0.252ms±0.121	0.223ms±0.086	0.126ms	~1.00x	≈	93.6%	50
findMany relation some	1.97ms±0.401	1.29ms±0.348	—	~1.00x	≈	71.5%	50
findMany relation every	15.08ms±4.43	10.50ms±1.19	—	1.44x	**	36.4%	50
findMany relation none	178.32ms±7.96	6.56ms±2.80	—	27.18x	***	45.7%	5
findMany nested relation	0.507ms±0.094	0.399ms±0.042	—	~1.00x	≈	34.8%	50
findMany complex	1.01ms±0.138	0.871ms±0.117	0.859ms	~1.00x	≈	43.3%	50
findFirst	0.080ms±0.033	0.034ms±0.030	0.070ms	~1.00x	≈	147.4%	50
findFirst skip	0.086ms±0.069	0.035ms±0.037	0.082ms	~1.00x	≈	154.8%	50
findUnique id	0.081ms±0.039	0.027ms±0.017	0.055ms	~1.00x	≈	115.3%	50
findUnique email	0.075ms±0.039	0.027ms±0.031	0.054ms	~1.00x	≈	202.7%	50
count	0.037ms±0.0072	0.016ms±0.0064	0.015ms	~1.00x	≈	90.1%	50
count where	0.427ms±0.070	0.278ms±0.052	0.253ms	~1.00x	≈	52.2%	50
_count via include	0.745ms±0.105	0.382ms±0.091	—	~1.00x	≈	58.8%	50
_count via include + relations	1.53ms±0.288	0.691ms±0.291	—	~1.00x	≈	105.5%	50
aggregate count	0.076ms±0.014	0.026ms±0.011	—	~1.00x	≈	112.6%	50
aggregate sum/avg	0.246ms±0.052	0.189ms±0.037	—	~1.00x	≈	55.2%	50
aggregate where	0.255ms±0.061	0.214ms±0.040	—	~1.00x	≈	52.2%	50
aggregate min/max	0.680ms±0.165	0.393ms±0.106	—	~1.00x	≈	69.7%	50
aggregate complete	0.568ms±0.292	0.391ms±0.095	—	~1.00x	≈	62.3%	50
groupBy	0.934ms±0.104	0.859ms±0.136	—	~1.00x	≈	50.0%	50
groupBy count	0.874ms±0.100	0.794ms±0.102	—	~1.00x	≈	41.1%	50
groupBy multi	1.74ms±0.119	1.53ms±0.141	—	~1.00x	≈	31.2%	50
groupBy having	0.843ms±0.131	0.747ms±0.098	—	~1.00x	≈	42.5%	50
groupBy + where	0.450ms±0.086	0.337ms±0.097	—	~1.00x	≈	70.0%	50
groupBy aggregates	1.09ms±0.124	0.917ms±0.116	—	~1.00x	≈	40.4%	50
groupBy min/max	1.18ms±0.096	0.890ms±0.095	—	~1.00x	≈	34.6%	50
include posts	1.45ms±0.231	0.892ms±0.190	2.55ms	~1.00x	≈	58.1%	50
include profile	0.147ms±0.028	0.071ms±0.017	0.202ms	~1.00x	≈	65.0%	50
include 3 levels	0.920ms±0.250	0.778ms±0.081	0.915ms	~1.00x	≈	33.8%	50
include 4 levels	0.799ms±0.085	0.465ms±0.049	0.670ms	~1.00x	≈	32.9%	50
include + where	0.810ms±0.401	0.571ms±0.173	0.508ms	~1.00x	≈	73.1%	50
include + select nested	0.645ms±0.153	0.591ms±0.074	1.20ms	~1.00x	≈	40.9%	50
findMany startsWith	0.071ms±0.016	0.041ms±0.0069	0.066ms	~1.00x	≈	52.4%	50
findMany endsWith	0.252ms±0.096	0.132ms±0.134	0.184ms	~1.00x	≈	184.3%	50
findMany NOT contains	0.296ms±0.112	0.102ms±0.085	0.192ms	~1.00x	≈	145.2%	50
findMany LIKE	0.113ms±0.055	0.063ms±0.042	0.099ms	~1.00x	≈	106.0%	50
findMany <	7.95ms±0.626	2.95ms±0.193	5.36ms	2.70x1.82x D	***	21.8%	50
findMany <=	9.47ms±0.948	3.46ms±0.432	5.03ms	2.73x1.45x D	***	39.8%	50
findMany >	5.61ms±0.633	1.98ms±0.501	2.93ms	2.83x1.48x D	***	73.5%	50
findMany NOT IN	0.231ms±0.235	0.086ms±0.026	0.132ms	~1.00x	≈	70.0%	50
findMany isNot null	0.183ms±0.062	0.079ms±0.035	0.115ms	~1.00x	≈	89.4%	50
orderBy multi-field	0.811ms±0.280	0.678ms±0.114	0.790ms	~1.00x	≈	48.5%	50
orderBy relation to-one	0.628ms±0.173	0.517ms±0.128	—	~1.00x	≈	63.8%	50
orderBy relation + scalar	0.550ms±0.063	0.443ms±0.027	—	~1.00x	≈	20.6%	50
orderBy nullable relation	0.455ms±0.088	0.385ms±0.060	—	~1.00x	≈	46.8%	50
orderBy deep relation	0.500ms±0.083	0.434ms±0.055	—	~1.00x	≈	37.9%	50
orderBy deep relation + scalar	0.937ms±0.108	0.767ms±0.088	—	~1.00x	≈	36.3%	50
distinct status	8.58ms±0.760	5.89ms±0.577	—	1.46x	***	29.9%	50
distinct multi	9.94ms±0.660	7.19ms±0.539	—	1.38x	***	24.1%	50
select + include	0.497ms±0.076	0.125ms±0.034	0.202ms	~1.00x	≈	72.4%	50
_count relation	0.725ms±0.129	0.533ms±0.111	—	~1.00x	≈	58.3%	50
_count multi-relation	0.103ms±0.108	0.055ms±0.034	—	~1.00x	≈	107.3%	50
_count inside include	0.963ms±0.118	0.501ms±0.096	—	~1.00x	≈	57.7%	50
_count inside nested select	1.54ms±0.156	0.840ms±0.095	—	~1.00x	≈	36.7%	50
_count deep include	1.39ms±0.118	0.976ms±0.066	—	~1.00x	≈	23.4%	50
findMany Date range	0.099ms±0.060	0.054ms±0.042	0.316ms	~1.00x	≈	128.0%	50
count Date range	0.478ms±0.157	0.410ms±0.109	0.338ms	~1.00x	≈	70.1%	50
findMany Date gte	0.091ms±0.0089	0.049ms±0.014	0.083ms	~1.00x	≈	76.3%	50
depth-1 low-fan Project→labels	0.123ms±0.017	0.069ms±0.019	—	~1.00x	≈	79.7%	50
depth-1 mid-fan Project→tasks	0.871ms±0.203	0.437ms±0.087	—	~1.00x	≈	54.0%	50
depth-1 high-fan User→assignedTasks	1.24ms±0.363	0.645ms±0.248	—	~1.00x	≈	94.6%	50
depth-1 wide	1.63ms±0.453	0.938ms±0.145	—	1.74x	***	50.3%	50
depth-1 no-limit Project→tasks	11.69ms±1.15	5.10ms±0.731	—	2.29x	***	40.4%	50
depth-2	0.906ms±0.194	0.544ms±0.050	—	~1.00x	≈	29.6%	50
depth-2 paginated	0.661ms±0.250	0.717ms±0.095	—	~1.00x	≈	40.5%	50
depth-2 high-fan	1.76ms±0.328	1.03ms±0.182	—	~1.00x	≈	51.6%	50
depth-2 wide+attach+activity	2.13ms±0.429	1.26ms±0.486	—	~1.00x	≈	95.4%	50
depth-2 no-limit	23.64ms±2.04	12.97ms±1.44	—	1.82x	***	38.2%	50
depth-3	5.23ms±0.329	2.79ms±0.696	—	1.88x	***	67.4%	50
depth-3 paginated	1.45ms±0.246	0.967ms±0.099	—	~1.00x	≈	33.0%	50
depth-4	3.22ms±0.295	1.78ms±0.430	—	1.81x	***	68.1%	50
depth-4 paginated	0.782ms±0.177	0.435ms±0.025	—	~1.00x	≈	19.1%	50
findFirst depth-2	0.538ms±0.060	0.330ms±0.053	—	~1.00x	≈	45.5%	50
findUnique depth-2	0.550ms±0.306	0.358ms±0.071	—	~1.00x	≈	57.1%	50

*** p < 0.001** p < 0.01* p < 0.05ns not significant≈ within 1ms noise floor± 95% CICV% > 15% = high variance

View raw benchmark data on GitHub →

Methodology

–Reproduce with: npm run bench (BENCHMARK_JSON_OUTPUT=1 writes JSON). Results land in benchmark-results/summary-.json; v--latest.json feeds this page.
–Every query runs against the same database for all clients: Prisma, prisma-sql, Drizzle. Per-test sample size is shown in the n column.
–Reported per side: mean, median, stdDev, p95, p99, CV%. Significance uses Welch's t-test with a 1ms practical threshold; error bars are 95% CI.
–Runtime mode measured: SQL is converted per request. Prerendered (@optimize) queries skip conversion entirely and are faster than shown.
–Hardware and versions vary between runs - each JSON file carries its own timestamp. Compare runs from the same machine.

What runs faster, and what stays in Prisma

Every Prisma Client feature we checked, on PostgreSQL and SQLite. Writes stay in Prisma; reads are compiled to SQL where the shape is supported.

PostgreSQL

SQLite

Model methodsQuery optionsFiltersOrderingAggregatesClient APIResult types

25 of 73 checked features are fully accelerated on PostgreSQL. 19 can return a different result than Prisma; the full map lists each one with the reason.

Feature count by status
Status	PostgreSQL	SQLite
Accelerated	25	21
Partial	6	6
Falls back	9	7
Runs in Prisma	7	7
Can differ	19	17
Not available	2	4
prisma-sql only	5	2
Not in Prisma	0	6
Not checked	0	3

Open the full parity map

ValidatedHow do we ensure correctness?Supported queries are validated against Prisma across PostgreSQL and SQLite, with benchmark reports covering Prisma v6, v7, v8, Drizzle, and generated SQL.View Testing Methodology

How It Optimizes Prisma

Bypass Prisma's read execution path while keeping Prisma's API and types

Intercept Prisma Queries

Extension catches read operations (findMany, findFirst, findUnique, count, aggregate, groupBy) before they execute

generated/sql, excerpt

query: {
  $allModels: {
    async $allOperations({ model, operation, args, query }) {
      if (!ACCELERATED_METHODS.has(operation)) {
        return query(args)
      }
      return executeAccelerated(model, operation, args, query)
    },
  },
}

Generate Optimized SQL

Convert Prisma queries into fast, parameterized SQL with optimized JOINs

SQL for PostgreSQL

SELECT user_t.id, user_t.email, user_t.status,
  COALESCE((
    SELECT COALESCE(json_agg(json_build_object(
      'id', posts_0.id,
      'title', posts_0.title,
      'authorId', posts_0."authorId"
    )), '[]'::json)
    FROM "public"."Post" posts_0
    WHERE posts_0."authorId" = user_t.id
  ), '[]'::json) AS posts
FROM "public"."User" user_t
WHERE user_t.status = $1

-- params ["ACTIVE"]

Execute Directly

Run queries through postgres.js or better-sqlite3, bypassing Prisma read overhead

generated/sql, excerpt

if (DIALECT === 'postgres') {
  await client.unsafe(sql, normalizedParams).forEach((row) => {
    results.push(row)
  })
  return results
}

const stmt = getOrPrepareStatement(client, sql)
return stmt.all(...normalizedParams)

Return Compatible Results

Results match Prisma's expected shape. Types, IntelliSense, and existing query code remain unchanged

users.ts

const users = await prisma.user.findMany({
  where: { status: 'ACTIVE' },
  include: { posts: true },
})

users[0].email
users[0].posts[0].title

// Same inferred types as Prisma Client.
// Decimal, BigInt and DateTime values can
// differ; the parity map lists each case.

generated/sql, excerpt

query: {
  $allModels: {
    async $allOperations({ model, operation, args, query }) {
      if (!ACCELERATED_METHODS.has(operation)) {
        return query(args)
      }
      return executeAccelerated(model, operation, args, query)
    },
  },
}

SELECT user_t.id, user_t.email, user_t.status,
  COALESCE((
    SELECT COALESCE(json_agg(json_build_object(
      'id', posts_0.id,
      'title', posts_0.title,
      'authorId', posts_0."authorId"
    )), '[]'::json)
    FROM "public"."Post" posts_0
    WHERE posts_0."authorId" = user_t.id
  ), '[]'::json) AS posts
FROM "public"."User" user_t
WHERE user_t.status = $1

-- params ["ACTIVE"]

if (DIALECT === 'postgres') {
  await client.unsafe(sql, normalizedParams).forEach((row) => {
    results.push(row)
  })
  return results
}

const stmt = getOrPrepareStatement(client, sql)
return stmt.all(...normalizedParams)

const users = await prisma.user.findMany({
  where: { status: 'ACTIVE' },
  include: { posts: true },
})

users[0].email
users[0].posts[0].title

// Same inferred types as Prisma Client.
// Decimal, BigInt and DateTime values can
// differ; the parity map lists each case.

Verified SecureHow do we prevent SQL injection?Multi-layer validation with parameter binding, schema-based field checks, and the current security test suiteView Security Details

Why Choose This Prisma Extension?

Get raw SQL execution speed for reads while keeping Prisma's developer experience

Instant Prisma Speedup: One-time setup accelerates Prisma reads. No refactoring, no migration, no downtime.
Keep Your Prisma Types: Full TypeScript support maintained. Type inference, autocomplete, and compile-time safety preserved while accelerating reads.
Production-Tested Solution: Current parity tests validate supported reads across recent Prisma versions. Use it for production apps after checking compatibility with your Prisma version and workload.
Multiple Database Support: Optimize Prisma reads on PostgreSQL (including Neon, Supabase) and SQLite.
Pre-Compiled Option: Optional generator creates build-time SQL, reducing overhead to microseconds for your hottest queries.
Serverless Ready (Node Runtimes): Works in serverless Node runtimes. Edge runtime support depends on runtime constraints and the driver you use.

Beyond faster reads

prisma-sql adds capabilities Prisma Client does not have. Same types, same code style.

$batch

N dashboard queries, one round trip

Combine independent reads into a single PostgreSQL round trip. Typed object result, no Promise.all juggling.

const dash = await prisma.$batch((b) => ({
  users: b.user.count(),
  posts: b.post.findMany({ take: 10 }),
  stats: b.task.aggregate({ _count: true }),
}))

@optimize

Prebake your hottest queries

Annotate a model in schema.prisma. At generate time the exact SQL is baked; matching runtime shapes skip conversion completely.

/// @optimize {
///   "method": "findMany",
///   "query": {
///     "where": { "status": "$status" },
///     "take": "$take"
///   }
/// }
model User { ... }

Cardinality planner

Strategy chosen by data, not by guess

Collected relation stats drive strategy switching between flat joins, correlated subqueries and where-in segmented loading. Large child tables never trigger per-parent scans.

prisma-sql-collect-stats \
  --output dist/planner.generated.js \
  --prisma-client dist/client/index.js

When Prisma Performance Matters Most

Common scenarios where this extension makes a real difference

Analytics & Reporting

Prisma aggregations and groupBy operations benefit significantly from direct SQL execution

5× faster groupBy accelerates reports
Faster aggregate queries
Real-time metrics with lower latency

High-Traffic APIs

Per-query overhead compounds under load, especially on read-heavy endpoints

Lower API response times
Handle more requests per instance
Reduce infrastructure costs

Serverless Functions

Every millisecond matters in serverless: reduce read latency where it counts

Better p95/p99 on reads
Lower costs through faster reads
Faster reads without refactors

Mobile Backends

Users notice latency: faster reads improve perceived UX immediately

Faster feed loading
Faster pagination
More responsive interactions

Optimize Prisma in 3 Steps

Accelerate Prisma reads in under 60 seconds

Install the Prisma Extension

Terminal

# PostgreSQL
npm install prisma-sql postgres

# SQLite
npm install prisma-sql better-sqlite3

Add Extension to Prisma Client

schema.prisma

generator client {
  provider = "prisma-client"
}

generator sql {
  provider = "prisma-sql-generator"
}

Terminal

npx prisma generate

db.ts

import { PrismaClient } from '@prisma/client'
import { speedExtension, type SpeedClient } from './generated/sql'
import postgres from 'postgres'

const sql = postgres(process.env.DATABASE_URL!)
const basePrisma = new PrismaClient()

export const prisma = basePrisma.$extends(
  speedExtension({ postgres: sql }),
) as SpeedClient<typeof basePrisma>

Use Prisma Normally

users.ts

const users = await prisma.user.findMany({
  where: { status: 'ACTIVE' },
  include: { posts: true },
})

SQLite: pass speedExtension({ sqlite: db }) with a better-sqlite3 database.Full setup guide

Prisma Performance FAQ

Common questions about optimizing Prisma reads

Why does Prisma have overhead?

Prisma adds overhead because it implements API guarantees like schema-based validation, consistent query behavior, and result shaping. Those layers provide a great developer experience but cost time compared to executing raw SQL directly.

How do I optimize Prisma queries?

Optimize read-heavy Prisma workloads by adding this extension. It executes read operations via direct SQL using postgres.js or better-sqlite3 while keeping Prisma's API and types. Setup is a small initialization change and does not require refactoring your existing queries.

Is Prisma slower than raw SQL?

For many read workloads, yes. There is architectural overhead compared to raw SQL execution. This extension aims to keep Prisma's DX while reducing read latency by executing SQL directly.

Can I speed up Prisma without changing existing queries?

Yes. Add the extension once during Prisma Client initialization and keep your existing Prisma query code unchanged. Read operations run faster while your Prisma API, types, and schema remain the same.

Does this work in production?

Yes, it is designed for production use on supported read paths. Always verify compatibility with your Prisma version and run your own regression tests before rollout.

What causes slower Prisma aggregations?

Aggregations and groupBy often amplify fixed overhead (query processing and result shaping) and can involve larger intermediate result sets. This extension optimizes those reads by generating SQL directly, which typically reduces latency on aggregation-heavy endpoints.

Ready to Speed Up Prisma?

Check the current benchmark table for supported PostgreSQL and SQLite reads

Install Prisma-SQL View on GitHub

Fix Slow Prisma Queries With Direct SQL Execution

Why Is Prisma Slow?

The Prisma Journey: DX First, Performance Matters at Scale

💡 Why This Extension Exists

🎯 2019: The Birth of Modern Prisma

⚡ 2020–2022: Rapid Growth & Feature Expansion

📊 2023: Overhead Becomes Noticeable at Scale

🚀 2024–2025: Prisma Performance Work Continues

🎯 2026: prisma-sql Extension Released

Query Translation Layer

Validation & Type Guarantees

Result Shaping

How Much Faster? Real Benchmarks

Detailed Benchmark Results

PostgreSQLPrisma v6 (6.19.3)

PostgreSQLPrisma v7 (7.10.0)

PostgreSQLPrisma v8 (8.1.0-dev.1)

SQLitePrisma v6 (6.19.3)

SQLitePrisma v7 (7.10.0)

SQLitePrisma v8 (8.1.0-dev.1)

Methodology

What runs faster, and what stays in Prisma

How It Optimizes Prisma

Intercept Prisma Queries

Generate Optimized SQL

Execute Directly

Return Compatible Results

Why Choose This Prisma Extension?

Beyond faster reads

$batch

@optimize

Cardinality planner

When Prisma Performance Matters Most

Analytics & Reporting

High-Traffic APIs

Serverless Functions

Mobile Backends

Optimize Prisma in 3 Steps

Install the Prisma Extension

Add Extension to Prisma Client

Use Prisma Normally

Prisma Performance FAQ

Ready to Speed Up Prisma?