turboapi 0.5.22

FastAPI-compatible web framework with Rust HTTP core - 2-3x faster with Python 3.13 free-threading

TurboAPI

The FastAPI you know. The speed you deserve.

Meet Turbito — the tiny Rust-powered engine that makes your FastAPI fly.

The Problem • The Solution • Meet Turbito • Quick Start • What's New • Benchmarks

---

The Problem

You love FastAPI. The clean syntax. The automatic validation. The beautiful docs. But then you deploy to production, and the reality hits:

"Why is my simple API only handling 8,000 requests per second?"

You've optimized your database queries. Added caching. Switched to async. Still not fast enough. The bottleneck isn't your code—it's the framework itself.

Python's GIL (Global Interpreter Lock) means only one thread executes Python code at a time. JSON serialization happens in pure Python. HTTP parsing happens in pure Python. Every microsecond adds up.

The Solution

TurboAPI is FastAPI with a Rust-powered engine. Same API. Same syntax. 1.3-1.8x faster.

# This is all you change
from turboapi import TurboAPI as FastAPI

Everything else stays exactly the same.

Why It's Faster

What FastAPI Does	What TurboAPI Does	Speedup
HTTP parsing in Python	HTTP parsing in Rust (Hyper/Tokio)	3x
JSON with json.dumps()	SIMD-accelerated JSON (simd-json)	2x
GIL-bound threading	Python 3.13 free-threading	2x
dict-based routing	Radix tree with O(log n) lookup	1.5x
Async via asyncio	Async via Tokio work-stealing	1.2x

The result? Your existing FastAPI code runs faster without changing a single line of business logic.

---

Meet Turbito

Turbito is the little engine inside TurboAPI.

While you write normal FastAPI code, Turbito is:

• Parsing HTTP in Rust (Hyper/Tokio)
• Serializing JSON with SIMD acceleration
• Scheduling async work with Tokio's work-stealing scheduler
• Dodging the GIL like a speed demon

You never see Turbito. You just feel the speed.

---

What's New

v0.5.21 — Free-Threading Stability Release

This release fixes critical issues when running with free-threaded Python (3.13t) and Metal/MLX GPU frameworks:

Fix	Description
Memory Corruption	Fixed race condition where request body bytes were corrupted when using async handlers with MLX models loaded
Response Serialization	Response objects now properly serialize their content instead of string representation
Async BaseModel	Async handlers with BaseModel parameters now correctly receive the validated model instance
JSON Parsing	Added Python fallback for edge cases where simd-json is too strict

Technical Deep Dive: When running free-threaded Python with Metal GPU frameworks, memory can be accessed concurrently by the CPU and GPU. We now use defensive copying (PyBytes::new() in Rust, bytes(bytearray()) in Python) to ensure request data is isolated before processing.

# Upgrade to get the fixes
pip install --upgrade turboapi

---

Quick Start

Installation

pip install turboapi

Requirements: Python 3.13+ (free-threading recommended for best performance)

Hello World

from turboapi import TurboAPI

app = TurboAPI()

@app.get("/")
def hello():
    return {"message": "Hello World"}

app.run()

That's it. Your first TurboAPI server is running at http://localhost:8000.

Async Handlers (New!)

TurboAPI now supports true async with Tokio-powered execution:

from turboapi import TurboAPI
import asyncio

app = TurboAPI()

@app.get("/sync")
def sync_handler():
    return {"type": "sync", "message": "Fast!"}

@app.get("/async")
async def async_handler():
    await asyncio.sleep(0.001)  # Simulated I/O
    return {"type": "async", "message": "Even faster under load!"}

app.run()

Async handlers are automatically detected and routed through Tokio's work-stealing scheduler for optimal concurrency.

Let Turbito Off the Leash

For maximum performance, run with Python's free-threading mode:

PYTHON_GIL=0 python app.py

This unlocks Turbito's full power by removing the GIL bottleneck. True parallelism, finally.

---

Benchmarks

Real numbers matter. Here's TurboAPI vs FastAPI on identical hardware:

Latest Benchmark Results

Endpoint	TurboAPI	FastAPI	Improvement
Sequential Latency
GET /	0.76ms	1.05ms	1.4x faster
GET /benchmark/simple	0.61ms	0.81ms	1.3x faster
GET /benchmark/medium	0.61ms	0.77ms	1.3x faster
GET /benchmark/json	0.72ms	1.04ms	1.4x faster
Concurrent Latency
GET /	2.05ms	2.53ms	1.2x faster
GET /benchmark/json	2.17ms	3.90ms	1.8x faster

Throughput (requests/second)

Endpoint	TurboAPI	FastAPI	Speedup
GET / (hello world)	19,596	8,336	2.4x
GET /json (object)	20,592	7,882	2.6x
GET /users/{id} (path params)	18,428	7,344	2.5x
POST /items (model validation)	19,255	6,312	3.1x

Async Handler Performance

Metric	Sync Handler	Async Handler	Notes
Sequential (100 req)	0.66ms	0.76ms	Similar performance
Concurrent (200 req)	108ms batch	139ms batch	Sync faster for CPU-bound
I/O Wait (1ms sleep)	2.22ms	2.06ms	Async wins for I/O

Key Insight: Use async handlers when you have actual I/O operations (database, network). For pure CPU work, sync handlers are slightly faster.

Run Your Own Benchmarks

# Quick benchmark
python benches/python_benchmark.py

# Full comparison with FastAPI
python tests/benchmark_comparison.py

# Async vs Sync comparison
python benches/async_comparison_bench.py

---

Async Support

How Async Works in TurboAPI

TurboAPI uses a hybrid async architecture:

┌─────────────────────────────────────────────────────────┐
│                Your Python Handlers                      │
│         @app.get("/sync")      @app.get("/async")       │
│         def handler():         async def handler():     │
├─────────────────────────────────────────────────────────┤
│              Handler Classification                      │
│   simple_sync │ body_sync │ simple_async │ body_async  │
├─────────────────────────────────────────────────────────┤
│              Tokio Runtime (Rust)                        │
│         Work-stealing scheduler • 14 workers            │
├─────────────────────────────────────────────────────────┤
│              pyo3-async-runtimes                         │
│     Python coroutines ↔ Rust futures conversion         │
└─────────────────────────────────────────────────────────┘

Handler Types

TurboAPI automatically classifies handlers for optimal dispatch:

Handler Type	Description	Use Case
simple_sync	Sync, no body	GET endpoints
body_sync	Sync, with body	POST/PUT without complex types
model_sync	Sync, with model validation	POST with dhi models
simple_async	Async, no body	GET with I/O operations
body_async	Async, with body	POST/PUT with I/O operations
enhanced	Full Python wrapper	Complex dependencies

When to Use Async

# Use sync for pure computation
@app.get("/compute")
def compute():
    result = sum(i * i for i in range(1000))
    return {"result": result}

# Use async for I/O operations
@app.get("/fetch-data")
async def fetch_data():
    async with aiohttp.ClientSession() as session:
        async with session.get("https://api.example.com/data") as resp:
            return await resp.json()

# Use async for database operations
@app.get("/users/{user_id}")
async def get_user(user_id: int):
    user = await database.fetch_one(query, values={"id": user_id})
    return {"user": user}

---

Migration Guide

TurboAPI is designed as a drop-in replacement for FastAPI. Here's how to migrate:

Step 1: Change Your Imports

# Before (FastAPI)
from fastapi import FastAPI, Depends, HTTPException, Query, Path
from fastapi.responses import JSONResponse, HTMLResponse
from fastapi.middleware.cors import CORSMiddleware

# After (TurboAPI)
from turboapi import TurboAPI as FastAPI, Depends, HTTPException, Query, Path
from turboapi.responses import JSONResponse, HTMLResponse
from turboapi.middleware import CORSMiddleware

Step 2: Update Your Models

TurboAPI uses dhi instead of Pydantic (it's API-compatible):

# Before (Pydantic)
from pydantic import BaseModel

# After (dhi)
from dhi import BaseModel

Step 3: Run Your App

# FastAPI way still works
if __name__ == "__main__":
    import uvicorn
    uvicorn.run(app, host="0.0.0.0", port=8000)

# Or use TurboAPI's built-in server (faster)
if __name__ == "__main__":
    app.run(host="0.0.0.0", port=8000)

That's it. Your FastAPI app is now a TurboAPI app.

---

Feature Parity

Everything you use in FastAPI works in TurboAPI:

Feature	Status	Notes
Route decorators (@get, @post, etc.)	✅	Full parity
Path parameters	✅	With type coercion
Query parameters	✅	With validation
Request body (JSON)	✅	SIMD-accelerated
Response models	✅	Full support
Async handlers	✅	Tokio-powered
Dependency injection	✅	Depends() with caching
OAuth2 authentication	✅	Password & AuthCode flows
HTTP Basic/Bearer auth	✅	Full implementation
API Key auth	✅	Header/Query/Cookie
CORS middleware	✅	Rust-accelerated
GZip middleware	✅	Configurable
Background tasks	✅	Async-compatible
WebSocket	✅	HTTP upgrade support
HTTP/2	✅	With server push
APIRouter	✅	Prefixes and tags
HTTPException	✅	With custom headers
Custom responses	✅	JSON, HTML, Redirect, etc.

---

Real-World Examples

API with Authentication

from turboapi import TurboAPI, Depends, HTTPException
from turboapi.security import OAuth2PasswordBearer

app = TurboAPI(title="My API", version="1.0.0")
oauth2_scheme = OAuth2PasswordBearer(tokenUrl="token")

@app.get("/users/me")
def get_current_user(token: str = Depends(oauth2_scheme)):
    if token != "secret-token":
        raise HTTPException(status_code=401, detail="Invalid token")
    return {"user": "authenticated", "token": token}

Async Database Access

from turboapi import TurboAPI
import asyncpg

app = TurboAPI()
pool = None

@app.on_event("startup")
async def startup():
    global pool
    pool = await asyncpg.create_pool("postgresql://localhost/mydb")

@app.get("/users/{user_id}")
async def get_user(user_id: int):
    async with pool.acquire() as conn:
        row = await conn.fetchrow("SELECT * FROM users WHERE id = $1", user_id)
        return dict(row) if row else {"error": "Not found"}

Request Validation

from dhi import BaseModel, Field
from typing import Optional

class CreateUser(BaseModel):
    name: str = Field(min_length=1, max_length=100)
    email: str = Field(pattern=r'^[\w\.-]+@[\w\.-]+\.\w+$')
    age: Optional[int] = Field(default=None, ge=0, le=150)

@app.post("/users")
def create_user(user: CreateUser):
    return {"created": True, "user": user.model_dump()}

CORS and Middleware

from turboapi.middleware import CORSMiddleware, GZipMiddleware

app.add_middleware(
    CORSMiddleware,
    allow_origins=["https://yourapp.com"],
    allow_methods=["*"],
    allow_headers=["*"],
)

app.add_middleware(GZipMiddleware, minimum_size=1000)

---

Architecture

TurboAPI's secret is a hybrid architecture where Python meets Rust:

┌──────────────────────────────────────────────────────────┐
│              Your Python Application                      │
│           (exactly like FastAPI code)                     │
├──────────────────────────────────────────────────────────┤
│         TurboAPI (FastAPI-compatible layer)              │
│      Routing • Validation • Dependency Injection          │
├──────────────────────────────────────────────────────────┤
│           Handler Classification (Phase 3+4)             │
│   simple_sync │ body_sync │ simple_async │ body_async   │
├──────────────────────────────────────────────────────────┤
│            PyO3 Bridge (zero-copy)                       │
│       Rust ↔ Python with minimal overhead                 │
├──────────────────────────────────────────────────────────┤
│            TurboNet (Rust HTTP Core)                     │
│   • Hyper + Tokio async runtime (14 worker threads)     │
│   • SIMD-accelerated JSON (simd-json)                    │
│   • Radix tree routing                                   │
│   • Zero-copy response buffers                           │
│   • pyo3-async-runtimes for async handler support       │
└──────────────────────────────────────────────────────────┘

Python handles the logic you care about. Routes, validation rules, business logic—all in Python.

Rust handles the heavy lifting. HTTP parsing, JSON serialization, connection management—the parts that need to be fast.

The result: FastAPI's developer experience with systems-level performance.

---

Building from Source

Want to contribute or build from source?

git clone https://github.com/justrach/turboAPI.git
cd turboAPI

# Create venv with Python 3.13 free-threading
python3.13t -m venv venv
source venv/bin/activate

# Build the Rust extension
pip install maturin
maturin develop --release

# Install Python package
pip install -e ./python

# Run tests
PYTHON_GIL=0 python -m pytest tests/ -v

# Run benchmarks
python benches/python_benchmark.py
python tests/benchmark_comparison.py

---

Roadmap

Completed ✅

• Rust HTTP core (Hyper/Tokio)
• SIMD JSON serialization & parsing
• Python 3.13 free-threading support
• FastAPI feature parity (OAuth2, Depends, Middleware)
• Radix tree routing with path parameters
• Handler classification for optimized fast paths
• Async handler optimization (Tokio + pyo3-async-runtimes)
• WebSocket HTTP upgrade support
• HTTP/2 with server push

In Progress 🚧

• OpenAPI/Swagger auto-generation

Planned 📋

• GraphQL support
• Database connection pooling
• Prometheus metrics
• Distributed tracing
• gRPC support

---

Community

• Issues & Features: GitHub Issues
• Discussions: GitHub Discussions
• Documentation:

---

License

MIT License. Use it, modify it, ship it.

---

Built for developers who love FastAPI.
Powered by Turbito ⚡

pip install turboapi