postgres-mysql.md

Results: PostgreSQL and MySQL

These are older results of this benchmark for the server databases PostgreSQL and MySQL; the main focus is now on embedded databases (see README.md). They were measured on a different machine (Python 3.14, macOS, Apple Silicon, 100 iterations per operation) with an earlier version of the benchmark, so don't compare them with the SQLite/ObjectBox results:

• They time creating the objects together with saving them; the current benchmark creates the objects before the timer starts and settles the GC (see Operations).
• Sync SQLAlchemy ("SA ORM") and sync SQLModel were added later and are not included; "SQLModel" in the charts refers to SQLModel async.

Operations and test models are described in the README.

Drivers:

ORM	Type	PostgreSQL	MySQL
Tortoise ORM	async	asyncpg	asyncmy
Django	sync	psycopg2	mysqlclient
peewee	sync	psycopg2	pymysql
SQLAlchemy ORM (async)	async	asyncpg	aiomysql
SQLObject	sync	psycopg2	mysqlclient
Piccolo	async	asyncpg	—
ormar	async	asyncpg	aiomysql
SQLModel (async)	async	asyncpg	aiomysql

Piccolo does not support MySQL. ObjectBox is an embedded database and is not part of these runs.

---

Results: PostgreSQL 17

PostgreSQL 17 in Docker. Driver: asyncpg for async ORMs, psycopg2 for sync ORMs.

Overview

Test 1: Simple Model (4 fields)

Operation	Django	peewee	SA ORM async	SQLObject	Tortoise ORM	Piccolo	ormar	SQLModel async	Best
Insert: Single	776	651	584	699	3,455	3,240	283	540	Tortoise ORM
Insert: Batch	2,069	2,418	1,154	1,575	14,033	12,138	1,048	1,081	Tortoise ORM
Insert: Bulk	5,224	5,009	1,264	—	18,807	18,060	1,178	1,079	Tortoise ORM
Filter: Large	57,451	62,009	32,318	50,694	143,810	147,031	19,735	32,389	Piccolo
Filter: Small	19,491	18,066	7,532	24,844	60,699	45,805	5,501	6,444	Tortoise ORM
Get	2,183	2,590	1,894	3,341	6,420	6,854	846	1,792	Piccolo
Filter: dict	64,394	81,652	33,798	—	209,380	328,124	26,777	31,117	Piccolo
Filter: tuple	55,080	66,827	37,933	—	197,175	332,310	23,286	34,411	Piccolo
Update: Whole	2,896	3,555	1,743	1,995	17,407	13,491	2,538	1,781	Tortoise ORM
Update: Partial	3,281	4,812	1,526	3,802	18,255	16,180	2,740	1,856	Tortoise ORM
Delete	3,386	5,701	1,869	843	21,010	18,968	2,890	1,895	Tortoise ORM
Geometric Mean	7,108	8,257	3,816	3,622	29,381	29,514	3,233	3,660	Piccolo

Test 2: FK Relations

Operation	Django	peewee	SA ORM async	SQLObject	Tortoise ORM	Piccolo	ormar	SQLModel async	Best
Insert: Single	765	822	564	517	3,350	3,075	264	521	Tortoise ORM
Insert: Batch	2,654	2,477	1,184	1,140	12,355	8,785	1,067	1,061	Tortoise ORM
Insert: Bulk	4,733	5,863	1,146	—	21,969	12,461	1,157	1,120	Tortoise ORM
Filter: Large	59,181	58,026	32,641	47,745	137,275	115,754	14,400	32,614	Tortoise ORM
Filter: Small	16,460	19,514	6,792	29,200	47,951	50,408	4,899	8,051	Piccolo
Get	2,394	2,605	1,905	1,440	5,870	5,665	787	1,922	Tortoise ORM
Filter: dict	73,174	78,313	33,632	—	207,239	303,737	25,848	31,831	Piccolo
Filter: tuple	65,459	70,256	34,130	—	192,246	284,733	29,597	36,273	Piccolo
Update: Whole	3,216	3,791	1,598	1,907	16,647	12,217	2,400	1,665	Tortoise ORM
Update: Partial	3,392	4,586	1,729	4,022	18,796	15,065	2,442	1,781	Tortoise ORM
Delete	866	5,880	896	629	17,921	12,984	2,940	1,711	Tortoise ORM
Geometric Mean	6,587	8,631	3,484	2,949	27,860	24,918	3,082	3,710	Tortoise ORM

Test 3: Wide Model (32+ fields)

Operation	Django	peewee	SA ORM async	SQLObject	Tortoise ORM	Piccolo	ormar	SQLModel async	Best
Insert: Single	660	712	555	683	1,560	2,761	177	507	Piccolo
Insert: Batch	1,580	1,338	1,137	1,280	8,077	4,627	542	953	Tortoise ORM
Insert: Bulk	3,374	3,074	1,187	—	9,571	5,596	814	971	Tortoise ORM
Filter: Large	32,188	25,686	26,031	26,710	50,995	29,431	2,893	23,445	Tortoise ORM
Filter: Small	8,946	9,546	7,167	20,497	31,247	18,652	2,105	5,997	Tortoise ORM
Get	1,541	1,311	1,773	2,288	5,056	3,563	626	1,708	Tortoise ORM
Filter: dict	28,900	32,299	22,351	—	71,887	121,813	6,883	22,634	Piccolo
Filter: tuple	34,619	30,980	29,794	—	66,990	121,798	6,353	29,111	Piccolo
Update: Whole	2,067	1,386	1,620	2,150	12,590	5,181	1,375	1,559	Tortoise ORM
Update: Partial	3,113	3,960	1,689	3,873	19,986	5,512	59	1,680	Tortoise ORM
Delete	3,480	4,915	1,728	769	22,214	14,305	2,595	1,574	Tortoise ORM
Geometric Mean	4,851	4,712	3,432	3,025	16,587	12,145	1,097	3,152	Tortoise ORM

---

Results: MySQL 8

MySQL 8 in Docker. Driver: asyncmy for Tortoise, aiomysql for SA async / ormar, mysqlclient for Django, pymysql for peewee / SQLObject. Piccolo does not support MySQL.

Overview

Test 1: Simple Model (4 fields)

Operation	Django	peewee	SA ORM async	SQLObject	Tortoise ORM	ormar	SQLModel async	Best
Insert: Single	460	475	1,260	386	2,226	1,218	1,198	Tortoise ORM
Insert: Batch	1,685	3,403	4,425	1,743	10,352	3,903	3,639	Tortoise ORM
Insert: Bulk	2,888	8,244	4,704	—	13,545	5,669	3,614	Tortoise ORM
Filter: Large	53,093	57,706	70,552	49,274	98,361	29,059	62,415	Tortoise ORM
Filter: Small	13,216	19,768	31,251	23,637	52,596	16,702	33,064	Tortoise ORM
Get	2,058	2,065	4,486	3,340	6,891	1,753	4,132	Tortoise ORM
Filter: dict	65,185	64,735	77,599	—	126,411	67,878	65,702	Tortoise ORM
Filter: tuple	62,068	57,817	101,337	—	124,069	69,064	99,444	Tortoise ORM
Update: Whole	2,422	3,203	2,376	2,211	12,485	5,086	2,152	Tortoise ORM
Update: Partial	3,568	4,624	2,510	6,088	14,040	6,045	2,439	Tortoise ORM
Delete	3,421	4,937	2,574	406	13,740	8,243	2,515	Tortoise ORM
Geometric Mean	6,035	8,000	8,998	3,307	21,351	8,885	8,221	Tortoise ORM

Test 2: FK Relations

Operation	Django	peewee	SA ORM async	SQLObject	Tortoise ORM	ormar	SQLModel async	Best
Insert: Single	477	525	1,173	399	1,777	952	1,063	Tortoise ORM
Insert: Batch	2,458	3,374	4,873	1,825	8,299	3,278	3,853	Tortoise ORM
Insert: Bulk	4,029	12,401	3,807	—	7,852	5,261	3,452	peewee
Filter: Large	60,201	50,582	71,476	46,076	95,298	21,057	69,532	Tortoise ORM
Filter: Small	13,404	14,109	26,652	21,838	58,097	8,354	31,293	Tortoise ORM
Get	2,146	2,171	3,965	3,227	7,188	1,719	4,333	Tortoise ORM
Filter: dict	70,525	63,725	70,878	—	122,091	48,115	54,117	Tortoise ORM
Filter: tuple	59,955	42,729	102,362	—	120,419	41,905	103,458	Tortoise ORM
Update: Whole	2,443	3,229	2,026	2,140	7,588	4,045	2,187	Tortoise ORM
Update: Partial	3,774	3,953	2,183	4,642	11,806	5,555	2,565	Tortoise ORM
Delete	917	5,191	883	328	12,982	5,491	2,461	Tortoise ORM
Geometric Mean	5,883	7,759	7,567	3,061	18,337	6,707	8,134	Tortoise ORM

Test 3: Wide Model (32+ fields)

Operation	Django	peewee	SA ORM async	SQLObject	Tortoise ORM	ormar	SQLModel async	Best
Insert: Single	368	366	1,318	342	1,331	703	1,071	Tortoise ORM
Insert: Batch	1,583	2,076	4,225	1,372	4,384	860	2,405	Tortoise ORM
Insert: Bulk	2,623	3,506	3,403	—	6,167	2,290	2,400	Tortoise ORM
Filter: Large	30,557	16,564	27,000	30,339	35,672	3,393	22,934	Tortoise ORM
Filter: Small	8,812	6,830	14,421	16,180	23,723	2,765	14,136	Tortoise ORM
Get	1,616	1,017	2,270	2,656	4,014	839	2,187	Tortoise ORM
Filter: dict	30,032	18,525	22,793	—	48,723	7,504	22,733	Tortoise ORM
Filter: tuple	35,392	21,983	31,740	—	41,779	7,659	29,487	Tortoise ORM
Update: Whole	1,852	1,590	1,456	2,207	8,684	1,715	1,410	Tortoise ORM
Update: Partial	3,407	4,148	1,512	5,091	13,981	5,052	1,517	Tortoise ORM
Delete	3,166	4,352	1,559	401	16,320	7,646	1,450	Tortoise ORM
Geometric Mean	4,466	3,943	4,928	2,692	11,560	2,629	4,292	Tortoise ORM

---

Analysis

PostgreSQL — Async ORMs dominate

Tortoise ORM and Piccolo share the top spots. Both use asyncpg, which provides binary protocol encoding and connection pooling — a massive advantage over psycopg2-based sync ORMs. Piccolo leads on simple reads (Filter: dict/tuple) thanks to its ultra-thin result mapping layer, while Tortoise excels at writes (Insert, Update, Delete) and wins overall on the more complex Test 2 and Test 3 schemas.

MySQL — Tortoise ORM leads decisively

Tortoise ORM wins every test by a wide margin (2x over the runner-up). The asyncmy driver (native async MySQL protocol) outperforms both aiomysql and sync mysqlclient/pymysql. SA ORM async is competitive on reads but falls behind on writes. Piccolo does not support MySQL.

Key takeaways

• For PostgreSQL/MySQL production workloads: Tortoise ORM and Piccolo (PG only) offer the best throughput
• SA ORM async / SQLModel async: Strong on reads but consistently slow on writes due to Session/UoW overhead

Running the benchmarks

Install as described in the README, then:

# PostgreSQL (needs psql reaching 127.0.0.1 as user postgres)
export DBTYPE=postgres PASSWORD=yourpassword PGPORT=5432
cd benchmarks && sh bench.sh

# MySQL (expects a Docker container named bench-mysql)
export DBTYPE=mysql PASSWORD=yourpassword MYPORT=3306
cd benchmarks && sh bench.sh

# PostgreSQL, MySQL and SQLite in one go
cd benchmarks && sh bench_all.sh