PostgreSQL REST API Benchmark 2026: 14 Frameworks Compared

Benchmark · Performance · January 2026

---

INFO

I wanted to repeat the benchmark from last year for several big reasons:

• First and foremost I wanted to make tests realistic and fair as much as possible. Last year tests were done in a hurry and to be honest, something was fisy with those results.
• Lately, there were some big internal changes in NpgsqlRest, many micro-optimizations and improvements that affected performance.
• I waanted to add more scenarios, and test different aspects (such as system resource usage which are now included).
• Some other frameworks also had major updates, so I wanted to see how they compare now.

Other than that, during the developement and explaration of the test project, I found some issues. For example:

• Pause between tests was way too short. Increaed from 5 to 30 seconds to allow TCP TIME_WAIT sockets to clear, JIT warmup, Garbage Collection, etc.
• Logging setup was uneven across frameworks, causing some to log to console/file during tests, impacting performance. Now all logging is disabled.

Other then that, I wanted to make the benchmarks as transparent, fair, realistic and accurate as possible. Everyone can check the source code used for tests, and run run themself if desired with their own setup. Also, anyone can suggest improvements or make PRs with new frameworks, optimizations or new test scenarios, and I would very much welcome that. As development of NpgsqlRest continues, I plan to repeat this benchmark from time to time to keep track of performance changes.

What follows is the detailed AI-generated analysis and report of the benchmark results. But you can also skip to summary results table or to conclusion sections directly.

Enjoy!

Following our 2025 benchmark, we've updated our testing with the latest framework versions and expanded our test scenarios. This year's benchmark includes higher concurrency levels (up to 200 VUs), pure HTTP overhead tests (Minimal Baseline), and POST body parsing benchmarks.

What We Tested

All frameworks executed the same PostgreSQL functions. This isolates the framework overhead from database performance - every framework runs identical queries against the same PostgreSQL instance. All functions use generate_series() for constant, predictable database response times with no table I/O.

Test Scenarios:

Scenario	Endpoint	Function	Description
Data Type Serialization	GET /api/perf-test	perf_test	Comprehensive test with 23 data types (text, int, numeric, bool, date/time, UUID, JSON, arrays). Returns 1, 10, 100, or 500 records.
Minimal Baseline (new)	GET /api/perf-minimal	perf_minimal	Pure HTTP routing overhead. Returns {"status": "ok", "ts": ...} - no parameters, minimal response. Tested at 100, 200, and 500 VUs.
POST Body Parsing (new)	POST /api/perf-post	perf_post	Tests JSON request body parsing. Sends nested JSON payload, echoes it back with computed values. Tested at 50 VUs with 10 and 100 records.
Nested JSON (new)	GET /api/perf-nested	perf_nested	Tests nested object serialization at configurable depth levels.
Large Payload (new)	GET /api/perf-large-payload	perf_large_payload	Tests chunked transfer and buffer handling with configurable KB-sized responses.
Many Parameters (new)	GET /api/perf-many-params	perf_many_params	Tests query string parsing with 20 parameters of mixed types.

Test Configuration:

• Load Tool: k6 load testing framework
• Duration: 60 seconds per test (30 seconds for minimal baseline)
• Concurrency: 1, 50, 100, and 200 virtual users (VUs) - up to 500 VUs for minimal baseline
• Environment: All services running in Docker containers on the same host
• Hardware: Hetzner Cloud CCX33 (General Purpose, x86 AMD) - 8 dedicated vCPUs, 32 GB RAM, 240 GB SSD, 30 TB traffic

Frameworks Tested:

Framework	Language	Version
NpgsqlRest (JIT)	.NET	3.4.7
NpgsqlRest (AOT)	.NET	3.4.7
PostgREST	Haskell	14.3
Go (net/http + pgx)	Go	1.25
Rust (Actix + tokio-postgres)	Rust	1.91.1
Spring Boot	Java 24	4.0.1
.NET Minimal API + Dapper	.NET 10	-
.NET Minimal API + EF Core	.NET 9/10	-
Fastify	Node.js	5.7.1
Bun	Bun	1.3.3
Swoole	PHP 8.4	6.0 (extension)
FastAPI	Python	0.128.0
Django	Python	6.0.1

What's New in This Benchmark

This benchmark introduces several improvements over the previous 2025 benchmark:

Version Updates

Framework	Previous Version	New Version	Changed
NpgsqlRest	3.2.2	3.4.7	Yes
PostgREST	12.2.8	14.3	Yes
Go	1.24	1.25	Yes
Rust	1.83.0	1.91.1	Yes
Bun	1.1.42	1.3.3	Yes
Fastify	5.6.2	5.7.1	Yes
FastAPI	0.127.1	0.128.0	Yes
Django	6.0	6.0.1	Yes
Swoole	PHP 8.4 + Swoole 6.0	PHP 8.4 + Swoole 6.0	No
Spring Boot	Java 24 + 4.0.1	Java 24 + 4.0.1	No
.NET Dapper	.NET 10 Preview	.NET 10 Preview	No
.NET EF Core	.NET 9 / .NET 10 Preview	.NET 9 / .NET 10 Preview	No

New Test Scenarios

We added five new benchmark scenarios (marked "Yes" in the table above):

• Minimal Baseline: Reveals true framework HTTP overhead separated from database I/O
• POST Body Parsing: Measures JSON request deserialization and response serialization
• Nested JSON: Tests nested object serialization performance
• Large Payload: Tests chunked transfer and buffer handling
• Many Parameters: Tests query string parsing with 20 parameters

Additionally, we extended concurrency testing:

• VU range expanded from 1/50/100 to 1/50/100/200
• Up to 500 VUs for minimal baseline tests
• Better reveals framework scaling limits under high load

Infrastructure Changes

• Resource Monitoring: New per-service CPU and memory tracking during tests
• JIT Warmup Phase: All frameworks now get a warmup period before benchmarks
• Extended Sleep Between Tests: 30 seconds between tests for TCP TIME_WAIT clearance
• Improved Result Aggregation: JSON output for programmatic analysis

Comparing With Previous Results

For detailed comparison with the 2025 benchmark:

• Previous benchmark branch: 202512280753
• Current benchmark branch: 202601211416
• Raw results: 202601211416_all.md

Key Findings

Swoole PHP Dominates Large Payload Scenarios

A major shift from 2025: Swoole PHP 6.0 now leads in most scenarios involving larger payloads. At 100 VU with 100 records, Swoole achieves 469.58 req/s - outperforming all competitors. With 500 records, Swoole maintains its lead at 106.88 req/s.

NpgsqlRest Leads High-Concurrency Low-Payload Scenarios

At 100 concurrent users with minimal payload (1 record), NpgsqlRest JIT achieves 4,588 req/s, maintaining its lead in low-latency scenarios. The gap between JIT and AOT versions has narrowed significantly in v3.4.7.

The Top Performers by Scenario

Scenario	Leader	Performance
1 VU, 1 Record	Bun	539.91 req/s
100 VU, 1 Record	NpgsqlRest JIT	4,588 req/s
100 VU, 100 Records	Swoole PHP	469.58 req/s
100 VU, 500 Records	Swoole PHP	106.88 req/s
Minimal Baseline (pure HTTP)	Go	20,104 req/s
POST Body Parsing	Go	9,628 req/s

Performance Tiers at 100 VU, 1 Record

Tier	Frameworks	Performance Range
Elite	NpgsqlRest JIT/AOT, Swoole PHP	4,400-4,600 req/s
Top	Bun, Go, Fastify, Spring Boot	4,100-4,400 req/s
High	.NET Dapper, Rust	3,900-4,100 req/s
Mid	.NET EF Core	3,400-3,500 req/s
Lower	FastAPI, PostgREST, Django	1,600-1,850 req/s

What Changed From 2025

Several notable performance shifts:

1. Swoole PHP improved dramatically - Jumping from mid-tier to top performer in data-heavy scenarios
2. Bun emerged as single-threaded champion - Best performance at 1 VU across all payload sizes
3. NpgsqlRest JIT/AOT gap narrowed - AOT now performs nearly identically to JIT in most scenarios
4. PostgREST improved - Better scaling under high concurrency compared to 2025
5. Go remains pure HTTP champion - Unmatched at 20,000+ req/s in minimal baseline tests

Scaling Behavior

Framework scaling patterns at increasing concurrency:

Framework	1 VU	50 VU	100 VU	200 VU	Scaling Factor
NpgsqlRest JIT	480	4,380	4,588	4,563	9.5x
Swoole PHP	471	4,160	4,423	4,485	9.5x
Bun	540	4,421	4,377	4,419	8.2x
PostgREST	271	1,818	1,749	1,663	6.5x

Large Payloads Level the Playing Field

With 500 records at 100 VU, database I/O dominates and the performance gap narrows:

Framework	500 Records @ 100 VU	Latency
Swoole PHP	106.88 req/s	468ms
Go	90.93 req/s	550ms
Rust	85.79 req/s	583ms
NpgsqlRest JIT	82.37 req/s	607ms
FastAPI	25.36 req/s	1,977ms

Pure HTTP Overhead (Minimal Baseline)

Testing pure HTTP handling without database access reveals framework overhead:

Framework	100 VU	200 VU	500 VU
Go	20,104 req/s	20,807 req/s	20,573 req/s
NpgsqlRest JIT	16,065 req/s	17,105 req/s	17,015 req/s
.NET Dapper	14,764 req/s	15,401 req/s	15,705 req/s
Spring Boot	14,138 req/s	14,593 req/s	14,435 req/s
Swoole PHP	12,042 req/s	12,309 req/s	12,297 req/s
PostgREST	5,410 req/s	4,974 req/s	5,324 req/s

Go's lightweight HTTP server achieves 20,000+ req/s - nearly 4x faster than PostgREST's pure HTTP overhead.

POST Body Parsing Performance

Testing JSON body parsing adds another dimension:

Framework	50 VU, 10 Records	50 VU, 100 Records
Go	9,629 req/s	2,697 req/s
Swoole PHP	7,470 req/s	2,445 req/s
Spring Boot	7,133 req/s	1,758 req/s
NpgsqlRest JIT	6,101 req/s	1,226 req/s
Bun	4,028 req/s	1,832 req/s
PostgREST	3,479 req/s	1,156 req/s

Go's efficient JSON parsing maintains its lead, while Bun shows strong performance with larger payloads relative to smaller ones.

Python Frameworks Continue to Struggle

Both FastAPI and Django remain at the bottom in most scenarios:

• FastAPI at 100 VU with 500 records: 1,977ms average latency
• Django performs better with data serialization but lags in pure throughput

JIT vs AOT in 2026

NpgsqlRest's JIT and AOT versions now perform nearly identically:

Scenario	JIT	AOT	Difference
100 VU, 1 Record	4,588 req/s	4,527 req/s	1.3%
100 VU, 100 Records	377 req/s	375 req/s	0.5%
Minimal Baseline	16,065 req/s	15,624 req/s	2.8%

For most workloads, the choice between JIT and AOT can now be based on deployment requirements (image size, cold start) rather than performance.

Why Certain Frameworks Excel

Swoole PHP's Rise

Swoole 6.0 brings significant improvements:

• Coroutine-based async I/O eliminates blocking
• Efficient memory management for large responses
• Native PostgreSQL driver optimization

Go's HTTP Dominance

Go's minimal baseline performance (20K+ req/s) demonstrates:

• Extremely low HTTP parsing overhead
• Efficient goroutine scheduling
• Zero-allocation hot paths in net/http

NpgsqlRest's Architecture

NpgsqlRest continues to excel by eliminating layers:

1. No ORM overhead - Direct PostgreSQL protocol via Npgsql
2. No routing framework - Endpoints derived from database metadata
3. No serialization layer - PostgreSQL handles JSON serialization
4. Efficient connection pooling - Npgsql's built-in pooling

Resource Usage

New in this benchmark: per-service memory and CPU monitoring during test execution.

Service	Peak Memory	Avg Memory	Avg CPU
Go	75.94 MB	21.35 MB	4.84%
Swoole PHP	78.99 MB	50.92 MB	4.49%
Rust	168.20 MB	43.03 MB	3.79%
Bun	168.00 MB	59.31 MB	4.29%
.NET Dapper	192.70 MB	96.55 MB	4.19%
FastAPI	198.10 MB	133.73 MB	4.29%
NpgsqlRest AOT	237.80 MB	59.84 MB	4.30%
.NET 10 EF	256.50 MB	143.59 MB	6.16%
.NET 9 EF	276.60 MB	133.80 MB	6.36%
NpgsqlRest JIT	321.30 MB	111.41 MB	4.42%
Fastify	411.40 MB	58.96 MB	3.68%
Django	824.00 MB	386.77 MB	11.84%
Spring Boot	1,010.00 MB	806.79 MB	5.38%

Key Observations

• Go has the lowest memory footprint (21 MB average, 76 MB peak)
• Swoole PHP achieves top-tier performance with minimal resources (51 MB average)
• NpgsqlRest AOT uses half the memory of JIT (60 MB vs 111 MB average)
• Spring Boot has the highest memory usage (807 MB average, 1 GB peak)
• Django has the highest CPU usage (11.84% average)

The full resource monitoring data is available in the stats directory.

Important Note: JSON and Array Type Handling

When interpreting these results, not all frameworks return identical JSON responses. The differences lie in how each framework handles PostgreSQL's JSON, JSONB, and array types.

Framework	json	jsonb	int[]	text[]
NpgsqlRest (JIT/AOT)	✅	✅	✅	✅
.NET EF Core / Dapper	✅	✅	✅	✅
Rust	✅	✅	✅	✅
Fastify	✅	✅	✅	✅
Django	✅	❌	✅	✅
Go	❌	❌	✅	✅
FastAPI	❌	❌	✅	✅
PostgREST	⚠️	⚠️	✅	✅
Bun	⚠️	⚠️	✅	✅
Spring Boot	⚠️	⚠️	✅	✅
Swoole PHP	❌	❌	❌	❌

✅ = Properly parsed as native JSON/array ❌ = Returns raw PostgreSQL text format (string) ⚠️ = Unusual format (wrapped in metadata or array)

Swoole PHP's impressive numbers come with a caveat: JSON/JSONB and array fields require additional client-side parsing.

Conclusion

The 2026 benchmark reveals a more nuanced performance landscape:

• For low-latency, high-concurrency APIs: NpgsqlRest JIT remains the top choice at 4,588 req/s
• For data-heavy workloads: Swoole PHP now leads with superior large-payload handling
• For pure HTTP performance: Go is unmatched at 20,000+ req/s
• For balanced workloads: Bun, Go, and Fastify offer excellent all-around performance

The "database as API" approach continues to deliver exceptional results. NpgsqlRest requires zero application code - just configuration - while delivering top-tier performance and correct PostgreSQL type handling.

Lines of Code Comparison

Performance isn't everything. Development time, maintainability, and code complexity matter too. Here's how much code each framework requires to implement the same API endpoints:

Framework	Lines of Code
PostgREST	14 (config only)
NpgsqlRest	21 (config only)
Fastify	100
.NET EF	116
Bun	133
FastAPI	136
Spring Boot	139
.NET Dapper	140
Django	203
Swoole PHP	216
Rust	291
Go	347

Go may deliver 20,000+ req/s in pure HTTP tests, but that comes at the cost of writing 347 lines of boilerplate code for basic CRUD operations. Every line of code is a potential bug, a maintenance burden, and development time spent.

NpgsqlRest and PostgREST take a different approach: define your API in the database, configure once, and let the framework handle the rest. You get top-tier performance (4,500+ req/s) with just 14-21 lines of configuration - no serialization code, no routing logic, no connection pool management.

The benchmark source code and detailed results are available in the pg_function_load_tests repository.

---

Full Benchmark Results

Summary Tables

All results in requests per second (req/s). Sorted by 100/1 performance.

Data Type Serialization

Function: perf_test

Framework	1/1	1/10	100/1	100/100	100/500
NpgsqlRest JIT	480	265	🥇 4,588	377	82
NpgsqlRest AOT	466	262	🥈 4,527	375	82
Swoole PHP	471	🥇 292	🥉 4,423	🥇 470	🥇 107
Bun	🥇 540	243	4,377	353	79
Go	🥉 484	🥈 289	4,362	🥈 406	🥈 91
Fastify	482	260	4,172	351	73
Spring Boot	464	233	4,147	282	61
.NET Dapper	425	244	4,101	331	72
Rust	🥈 507	🥉 285	3,940	🥉 388	🥉 86
.NET 10 EF	377	220	3,515	331	72
.NET 9 EF	362	223	3,425	330	72
FastAPI	478	220	1,843	111	25
PostgREST	271	181	1,749	342	79
Django	271	178	1,691	307	71

Column headers = VU/Records (e.g., 100/1 = 100 VU, 1 record). See detailed results.

New Scenarios

Framework	Min	POST	Nest	Large	Params
Go	🥇 20,104	🥇 9,629	🥇 3,756	🥇 1,618	🥇 16,100
NpgsqlRest JIT	🥈 16,065	6,101	3,061	1,096	🥈 11,504
NpgsqlRest AOT	🥉 15,624	6,065	🥉 3,073	919	🥉 11,221
.NET Dapper	14,764	6,089	2,989	1,397	10,702
Spring Boot	14,138	🥉 7,133	2,230	🥈 1,515	10,324
Swoole PHP	12,042	🥈 7,470	🥈 3,426	🥉 1,500	9,458
Rust	11,761	5,400	2,824	1,509	9,231
.NET 10 EF	9,385	4,879	2,615	1,314	7,452
.NET 9 EF	8,989	4,710	2,558	1,326	7,182
Fastify	8,899	4,631	2,921	1,248	7,711
Bun	7,803	4,028	2,287	1,235	3,116
PostgREST	5,410	3,479	1,912	908	3,960
FastAPI	4,090	2,487	971	1,492	2,080
Django	2,535	2,266	1,677	1,071	2,222

• Min — Minimal Baseline (100 VU) — perf_minimal
• POST — POST Body Parsing (50 VU, 10 records) — perf_post
• Nest — Nested JSON Serialization (50 VU, depth 1) — perf_nested
• Large — Large Payload (25 VU, 100KB) — perf_large_payload
• Params — Many Parameters (50 VU, 20 params) — perf_many_params

---

Results are grouped by concurrency level and payload size, sorted by requests per second (highest first).

Data Type Serialization Tests

1 Virtual User, 1 Record

Function: perf_test

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
bun-app-v1.3.3	539.91/s	1.84ms	32,395	summary	source
rust-app-v1.91.1	506.53/s	1.96ms	30,392	summary	source
go-app-v1.25	483.79/s	2.05ms	29,028	summary	source
fastify-app-v5.7.1	481.94/s	2.06ms	28,917	summary	source
npgsqlrest-jit-v3.4.7	480.28/s	2.07ms	28,817	summary	source
fastapi-app-v0.128.0	478.18/s	2.08ms	28,692	summary	source
swoole-php-app-v6.0	471.18/s	2.11ms	28,272	summary	source
npgsqlrest-aot-v3.4.7	466.06/s	2.13ms	27,964	summary	source
java24-spring-boot-v4.0.1	464.36/s	2.14ms	27,863	summary	source
net10-minapi-dapper-jit	425.45/s	2.34ms	25,527	summary	source
net10-minapi-ef-jit	376.73/s	2.64ms	22,604	summary	source
net9-minapi-ef-jit	361.52/s	2.75ms	21,692	summary	source
postgrest-v14.3	271.38/s	3.67ms	16,284	summary	source
django-app-v6.0.1	270.87/s	3.68ms	16,254	summary	source

1 Virtual User, 10 Records

Function: perf_test

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
swoole-php-app-v6.0	291.51/s	3.42ms	17,491	summary	source
go-app-v1.25	288.60/s	3.45ms	17,317	summary	source
rust-app-v1.91.1	284.94/s	3.50ms	17,098	summary	source
npgsqlrest-jit-v3.4.7	264.71/s	3.76ms	15,883	summary	source
npgsqlrest-aot-v3.4.7	262.37/s	3.80ms	15,743	summary	source
fastify-app-v5.7.1	259.77/s	3.84ms	15,587	summary	source
net10-minapi-dapper-jit	244.13/s	4.08ms	14,649	summary	source
bun-app-v1.3.3	243.23/s	4.10ms	14,594	summary	source
java24-spring-boot-v4.0.1	232.71/s	4.28ms	13,964	summary	source
net9-minapi-ef-jit	223.13/s	4.47ms	13,389	summary	source
fastapi-app-v0.128.0	220.40/s	4.52ms	13,225	summary	source
net10-minapi-ef-jit	219.98/s	4.53ms	13,200	summary	source
postgrest-v14.3	180.58/s	5.52ms	10,835	summary	source
django-app-v6.0.1	178.29/s	5.59ms	10,698	summary	source

100 Virtual Users, 1 Record

Function: perf_test

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
npgsqlrest-jit-v3.4.7	4,588.02/s	10.88ms	275,381	summary	source
npgsqlrest-aot-v3.4.7	4,526.64/s	11.02ms	271,720	summary	source
swoole-php-app-v6.0	4,423.22/s	11.29ms	265,603	summary	source
bun-app-v1.3.3	4,377.29/s	11.41ms	262,711	summary	source
go-app-v1.25	4,362.06/s	11.44ms	261,787	summary	source
fastify-app-v5.7.1	4,171.93/s	11.97ms	250,370	summary	source
java24-spring-boot-v4.0.1	4,146.96/s	12.03ms	248,882	summary	source
net10-minapi-dapper-jit	4,100.54/s	12.17ms	246,098	summary	source
rust-app-v1.91.1	3,939.83/s	12.67ms	236,565	summary	source
net10-minapi-ef-jit	3,515.31/s	14.20ms	210,977	summary	source
net9-minapi-ef-jit	3,424.67/s	14.60ms	205,781	summary	source
fastapi-app-v0.128.0	1,842.81/s	27.12ms	110,610	summary	source
postgrest-v14.3	1,749.07/s	28.58ms	105,038	summary	source
django-app-v6.0.1	1,690.72/s	29.56ms	101,545	summary	source

100 Virtual Users, 100 Records

Function: perf_test

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
swoole-php-app-v6.0	469.58/s	106.44ms	28,257	summary	source
go-app-v1.25	405.52/s	123.21ms	24,378	summary	source
rust-app-v1.91.1	387.70/s	128.79ms	23,313	summary	source
npgsqlrest-jit-v3.4.7	377.42/s	132.43ms	22,691	summary	source
npgsqlrest-aot-v3.4.7	374.57/s	133.40ms	22,519	summary	source
bun-app-v1.3.3	352.79/s	141.70ms	21,217	summary	source
fastify-app-v5.7.1	351.07/s	142.62ms	21,212	summary	source
postgrest-v14.3	342.26/s	145.93ms	20,578	summary	source
net10-minapi-ef-jit	331.31/s	151.40ms	19,994	summary	source
net10-minapi-dapper-jit	331.16/s	150.92ms	19,914	summary	source
net9-minapi-ef-jit	329.64/s	151.67ms	19,836	summary	source
django-app-v6.0.1	307.22/s	162.61ms	18,480	summary	source
java24-spring-boot-v4.0.1	281.56/s	177.62ms	16,949	summary	source
fastapi-app-v0.128.0	111.17/s	449.98ms	6,774	summary	source

100 Virtual Users, 500 Records

Function: perf_test

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
swoole-php-app-v6.0	106.88/s	468.22ms	6,459	summary	source
go-app-v1.25	90.93/s	550.42ms	5,511	summary	source
rust-app-v1.91.1	85.79/s	583.48ms	5,192	summary	source
npgsqlrest-jit-v3.4.7	82.37/s	606.96ms	4,991	summary	source
npgsqlrest-aot-v3.4.7	81.89/s	610.99ms	4,956	summary	source
postgrest-v14.3	78.59/s	636.29ms	4,753	summary	source
bun-app-v1.3.3	78.55/s	637.59ms	4,772	summary	source
fastify-app-v5.7.1	73.00/s	685.34ms	4,463	summary	source
net10-minapi-dapper-jit	72.42/s	691.80ms	4,391	summary	source
net9-minapi-ef-jit	72.00/s	694.45ms	4,355	summary	source
net10-minapi-ef-jit	71.98/s	694.12ms	4,363	summary	source
django-app-v6.0.1	71.16/s	702.86ms	4,302	summary	source
java24-spring-boot-v4.0.1	60.75/s	822.37ms	3,683	summary	source
fastapi-app-v0.128.0	25.36/s	1,977.30ms	1,613	summary	source

---

Minimal Baseline (Pure HTTP Overhead)

100 Virtual Users

Function: perf_minimal

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
go-app-v1.25	20,104.17/s	2.47ms	603,258	summary	source
npgsqlrest-jit-v3.4.7	16,064.93/s	3.10ms	481,990	summary	source
npgsqlrest-aot-v3.4.7	15,623.97/s	3.19ms	468,869	summary	source
net10-minapi-dapper-jit	14,763.56/s	3.37ms	442,970	summary	source
java24-spring-boot-v4.0.1	14,137.76/s	3.52ms	424,203	summary	source
swoole-php-app-v6.0	12,042.03/s	4.13ms	361,335	summary	source
rust-app-v1.91.1	11,760.56/s	4.22ms	352,878	summary	source
net10-minapi-ef-jit	9,384.73/s	5.31ms	281,613	summary	source
net9-minapi-ef-jit	8,988.95/s	5.55ms	269,715	summary	source
fastify-app-v5.7.1	8,899.19/s	5.61ms	267,058	summary	source
bun-app-v1.3.3	7,803.09/s	6.40ms	234,250	summary	source
postgrest-v14.3	5,410.48/s	9.23ms	162,402	summary	source
fastapi-app-v0.128.0	4,089.65/s	12.21ms	122,737	summary	source
django-app-v6.0.1	2,535.28/s	19.71ms	76,114	summary	source

---

POST Body Parsing

50 Virtual Users, 10 Records

Function: perf_post

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
go-app-v1.25	9,628.69/s	2.58ms	577,788	summary	source
swoole-php-app-v6.0	7,470.15/s	3.33ms	448,231	summary	source
java24-spring-boot-v4.0.1	7,132.74/s	3.49ms	427,988	summary	source
npgsqlrest-jit-v3.4.7	6,100.82/s	4.08ms	366,070	summary	source
net10-minapi-dapper-jit	6,088.58/s	4.09ms	365,437	summary	source
npgsqlrest-aot-v3.4.7	6,065.30/s	4.11ms	363,949	summary	source
rust-app-v1.91.1	5,399.96/s	4.61ms	324,130	summary	source
net10-minapi-ef-jit	4,879.06/s	5.11ms	292,761	summary	source
net9-minapi-ef-jit	4,710.24/s	5.29ms	282,651	summary	source
fastify-app-v5.7.1	4,631.21/s	5.39ms	277,906	summary	source
bun-app-v1.3.3	4,027.52/s	6.20ms	241,687	summary	source
postgrest-v14.3	3,478.73/s	7.17ms	208,759	summary	source
fastapi-app-v0.128.0	2,487.28/s	10.04ms	149,259	summary	source
django-app-v6.0.1	2,266.37/s	11.02ms	136,044	summary	source

---

Nested JSON Serialization

50 Virtual Users, Depth 1

Function: perf_nested

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
go-app-v1.25	3,756.41/s	6.64ms	225,426	summary	source
swoole-php-app-v6.0	3,426.34/s	7.27ms	205,620	summary	source
npgsqlrest-aot-v3.4.7	3,073.40/s	8.12ms	184,428	summary	source
npgsqlrest-jit-v3.4.7	3,060.64/s	8.15ms	183,675	summary	source
net10-minapi-dapper-jit	2,988.63/s	8.35ms	179,345	summary	source
fastify-app-v5.7.1	2,921.44/s	8.54ms	175,319	summary	source
rust-app-v1.91.1	2,824.19/s	8.83ms	169,476	summary	source
net10-minapi-ef-jit	2,615.08/s	9.54ms	156,947	summary	source
net9-minapi-ef-jit	2,557.86/s	9.76ms	153,501	summary	source
bun-app-v1.3.3	2,286.95/s	10.92ms	137,261	summary	source
java24-spring-boot-v4.0.1	2,229.81/s	11.19ms	133,816	summary	source
postgrest-v14.3	1,912.23/s	13.06ms	114,792	summary	source
django-app-v6.0.1	1,676.67/s	14.90ms	100,661	summary	source
fastapi-app-v0.128.0	970.57/s	25.74ms	58,262	summary	source

---

Large Payload

25 Virtual Users, 100KB Payload

Function: perf_large_payload

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
go-app-v1.25	1,618.26/s	7.70ms	97,107	summary	source
java24-spring-boot-v4.0.1	1,514.97/s	8.23ms	90,930	summary	source
rust-app-v1.91.1	1,508.80/s	8.27ms	90,539	summary	source
swoole-php-app-v6.0	1,500.22/s	8.31ms	90,027	summary	source
fastapi-app-v0.128.0	1,491.75/s	8.36ms	89,518	summary	source
net10-minapi-dapper-jit	1,396.93/s	8.93ms	83,826	summary	source
net9-minapi-ef-jit	1,326.18/s	9.41ms	79,583	summary	source
net10-minapi-ef-jit	1,313.73/s	9.50ms	78,847	summary	source
fastify-app-v5.7.1	1,248.43/s	10.00ms	74,941	summary	source
bun-app-v1.3.3	1,234.55/s	10.11ms	74,093	summary	source
npgsqlrest-jit-v3.4.7	1,095.93/s	11.39ms	65,765	summary	source
django-app-v6.0.1	1,070.77/s	11.66ms	64,270	summary	source
npgsqlrest-aot-v3.4.7	919.29/s	13.58ms	55,175	summary	source
postgrest-v14.3	907.83/s	13.75ms	54,499	summary	source

---

Many Parameters (20 params)

50 Virtual Users

Function: perf_many_params

Framework	Requests/s	Avg Latency	Total Requests	Summary	Source
go-app-v1.25	16,100.46/s	1.54ms	966,176	summary	source
npgsqlrest-jit-v3.4.7	11,503.53/s	2.16ms	690,265	summary	source
npgsqlrest-aot-v3.4.7	11,220.54/s	2.22ms	673,270	summary	source
net10-minapi-dapper-jit	10,701.77/s	2.33ms	642,149	summary	source
java24-spring-boot-v4.0.1	10,323.63/s	2.41ms	619,498	summary	source
swoole-php-app-v6.0	9,457.64/s	2.63ms	567,509	summary	source
rust-app-v1.91.1	9,230.96/s	2.69ms	553,984	summary	source
fastify-app-v5.7.1	7,710.72/s	3.23ms	462,679	summary	source
net10-minapi-ef-jit	7,452.38/s	3.34ms	447,165	summary	source
net9-minapi-ef-jit	7,181.52/s	3.47ms	430,916	summary	source
postgrest-v14.3	3,960.49/s	6.30ms	237,672	summary	source
bun-app-v1.3.3	3,116.21/s	8.01ms	187,044	summary	source
django-app-v6.0.1	2,222.19/s	11.24ms	133,359	summary	source
fastapi-app-v0.128.0	2,079.91/s	12.01ms	124,813	summary	source

More Blog Posts:
Custom Types & Multiset · Performance & High Availability · End-to-End Type Checking · Database-Level Security · Multiple Auth Schemes & RBAC · PostgreSQL BI Server · Secure Image Uploads · CSV & Excel Ingestion · Real-Time Chat with SSE · External API Calls · Reverse Proxy & AI Service · Zero to CRUD API · NpgsqlRest vs PostgREST vs Supabase

Get Started:
Quick Start Guide · Installation · Configuration Guide