# FastAPI

> Validation with Pydantic, and the best ecosystem for model glue: `ai` requests earn 15% more. 1.15× CPU, and only on an ASGI server. 10 MB.

## Trade-offs

| Axis | Score |
| --- | --- |
| Throughput | |
| Tail latency | |
| Memory | |
| Velocity | |
| Resilience | |
| Operability | |

## Runs on

- [Gunicorn](https://wiki.playquorum.dev/runtimes/gunicorn.md)
- [Gunicorn with gevent](https://wiki.playquorum.dev/runtimes/gevent.md)
- [Uvicorn ASGI](https://wiki.playquorum.dev/runtimes/uvicorn.md)
- [Free-threaded CPython](https://wiki.playquorum.dev/runtimes/python_ft.md)
- [PyPy](https://wiki.playquorum.dev/runtimes/pypy.md)

## Engineering terms

- [Model inference](https://wiki.playquorum.dev/glossary/model-inference.md)

## Facts

FastAPI
- **Language:** Python
- **Overhead:** 1.15 ms CPU
- **Velocity:** ×1.3
- **Memory:** +10 MB
- **Startup:** +0 s
- **ORM:** No

## See also

- [Runtimes and frameworks](https://wiki.playquorum.dev/runtimes.md)

Source: https://wiki.playquorum.dev/frameworks/fastapi