Deploying AI Studio Models as APIs

Your script runs on your laptop. Now, let's put it on the internet.

The Microservice Pattern

We usually wrap the Gemini logic in a lightweight HTTP server (FastAPI or Flask).

# main.py
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
import google.generativeai as genai
import os

app = FastAPI()

# Configure
genai.configure(api_key=os.environ["GEMINI_API_KEY"])
model = genai.GenerativeModel('gemini-1.5-flash')

class PromptRequest(BaseModel):
    text: str

@app.post("/generate")
async def generate(req: PromptRequest):
    try:
        response = model.generate_content(req.text)
        return {"result": response.text}
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

Why Middle-Layer?

Why not call Gemini directly from the Frontend?

1. Security: Keeps API Key hidden.
2. Rate Limiting: You can throttle users who spam.
3. Logging: You can save prompts to your DB.

Summary

Build a thin wrapper API. Use FastAPI for async support (handling multiple requests while waiting for Gemini).

In the next lesson, we discuss Hosting.