Why Rustango: an honest comparison
Rustango is a batteries-included web framework for Rust: declare a model once and get an ORM, migrations, an admin, a REST API and multi-tenancy out of it. This page sets out what it is good at, how it compares with the frameworks people usually reach for, the benchmark numbers, and where you should still pick something else.
The short version
- Choose Rustango when you want the productivity of a batteries-included framework with the speed, memory use and compile-time checking of Rust, and one binary to deploy.
- Choose something else when your team does not write Rust, when you need a huge package ecosystem today, or when you want a framework with a decade of production history.
What is in the box
| Area | Rustango |
|---|---|
| ORM | One model, correct SQL for PostgreSQL, MySQL 8+ and SQLite; relations, aggregates, soft delete |
| Migrations | Generated by comparing your models with the last snapshot, then applied |
| Admin | An auto-generated admin with search, filters, audit log and permissions |
| APIs | ViewSets and serializers for REST, OpenAPI generated from them |
| Auth | Sessions, JWT, OAuth2 / OIDC, single sign-on, API keys, HMAC, TOTP, passkeys |
| Multi-tenancy | Schema per tenant (PostgreSQL) or database per tenant (all three databases) |
| Background work | Jobs, scheduled tasks, signals, email, file storage |
| Security | CSRF, rate limits, login throttling, security headers, CORS, allowed hosts |
| AI agents | A built-in MCP server |
| Tooling | A project scaffolder, generators and one manage command line |
Everything beyond the ORM is a cargo feature, so a JSON-only API compiles the admin and templates out entirely.
How it compares
| Rustango | Django | Rails | Laravel | Loco | axum + sqlx | |
|---|---|---|---|---|---|---|
| Language | Rust | Python | Ruby | PHP | Rust | Rust |
| ORM | Built in | Built in | Built in | Built in | SeaORM | Queries by hand |
| Migrations from models | Yes, diffed | Yes, diffed | Written with generators | Written by hand | Written with generators | By hand |
| Admin | Built in | Built in | Gems | Packages (Nova, Filament) | No | No |
| REST API layer | Built in, with OpenAPI | Package (DRF) | Built in (API mode) | Built in | Built in | By hand |
| Multi-tenancy | Built in | Packages | Gems | Packages | No | By hand |
| Background jobs | Built in | Tasks API; worker via a package | Built in | Built in | Built in | By hand |
| Errors in models and queries | Caught at compile time | At run time | At run time | At run time | Caught at compile time | Caught at compile time |
| Deploy artefact | One binary | App + Python runtime | App + Ruby runtime | App + PHP runtime | One binary | One binary |
| Ecosystem | Small and new | Huge | Huge | Huge | Growing | Large (Rust crates) |
The honest reading: on features, Rustango is in the same class as Django and Rails, not ahead of them. It wins on speed, memory, compile-time safety and deployment, and against the other Rust options it wins on everything that comes in the box.
Benchmarks
Every app served the same blog: 1 000 posts, 10 000 comments, the same HTML checked byte for byte, 4 CPUs, 16 database connections, 50 concurrent connections. The full method, the limits and the profiles are in the benchmark write-up. Numbers from different rounds were measured separately, so each table stays within one round.
Round 2: compiled apps (requests per second, higher is better):
| App | Index | Detail | Tag | Index, cached | Memory under load |
|---|---|---|---|---|---|
| Rustango · actix | 17 401 | 27 292 | 7 056 | 231 975 | 6.3 MiB |
| Rustango · axum | 14 686 | 25 788 | 7 090 | 174 501 | 6.9 MiB |
| axum + sqlx | 14 140 | 22 842 | 6 467 | 198 669 | 7.6 MiB |
| actix + sqlx | 14 408 | 21 630 | 6 628 | 226 887 | 7.2 MiB |
| Rocket | 13 316 | 24 561 | 6 836 | 138 224 | 8.7 MiB |
| Loco | 10 290 | 15 635 | 5 604 | 153 243 | 12.0 MiB |
| Go (net/http + pgx) | 8 741 | 22 607 | 6 386 | 91 839 | 22.8 MiB |
Rustango, with its ORM, is the fastest of the eight on uncached pages, ahead of hand-written queries on raw sqlx.
Round 1: interpreted runtimes, with Rustango from the same run (the 0.60.6 release, before the round 2 fixes):
| App | Uncached index, req/s | Cached index, req/s | Memory under load |
|---|---|---|---|
| Rustango · actix | 6 341 | 200 167 | ~7 MiB |
| Laravel · Octane | 1 098 | 5 884 | 241 MiB |
| Python · gunicorn | 697 | 11 012 | 224 MiB |
| WordPress | 599 | 21 285 | 142 MiB |
| Laravel · php-fpm | 518 | 3 326 | 89 MiB |
The honest caveats
- Go is faster at pure computation, by 8–17% in the compute test, and plain axum with sqlx serves the cached index faster than Rustango on axum (198 669 against 174 501 req/s).
- It is young. Version 0.x, with a large surface still settling. Recent releases have added rather than removed, but read the changelog when you upgrade.
- The ecosystem is small. Django, Rails and Laravel have a package for almost anything; with Rustango you will more often write it yourself or reach for a general Rust crate.
- Rust has a learning curve, and compile times are real: minutes for a first build, seconds for a change.
- Schema-per-tenant needs PostgreSQL. On MySQL and SQLite, tenants get a database each.
- The benchmarks have limits. One laptop, a Docker VM, a read-heavy server-rendered workload. No writes, auth flows or heavy business logic. Your numbers will differ.
Who should use it
- Teams that like Django's or Rails' way of working but want Rust's speed, safety and small footprint.
- Rust teams tired of assembling the same auth, admin, migrations and API layers on every project.
- Multi-tenant SaaS products that want tenancy in the framework rather than bolted on.
Try it
cargo install cargo-rustango
cargo rustango new myapp
cd myapp && cargo run
The getting-started guide builds a blog with an admin, a JSON API and sign-in step by step. For content sites, Rustango-CMS is built on top of it.