import os
from fastapi import FastAPI, Request
from fastapi.responses import HTMLResponse
from openai import OpenAI
app = FastAPI()
MODEL = "nousresearch/hermes-2-pro-llama-3-8b"
def get_client():
key = os.getenv("NVIDIA_API_KEY")
if not key:
raise RuntimeError("❌ NVIDIA_API_KEY missing. Add it in Settings → Variables & secrets → Factory reboot.")
return OpenAI(base_url="https://integrate.api.nvidia.com/v1", api_key=key)
@app.get("/", response_class=HTMLResponse)
async def home():
return """
Hermes NIM
🟣 Hermes + NVIDIA NIM
Response will appear here...
"""
@app.post("/run")
async def run_hermes(req: Request):
try:
data = await req.json()
prompt = data.get("prompt", "")
if not prompt:
return {"error": "Missing prompt"}
client = get_client()
res = client.chat.completions.create(
model=MODEL,
messages=[{"role": "user", "content": prompt}],
temperature=0.7,
max_tokens=1024
)
return {"status": "done", "output": res.choices[0].message.content}
except Exception as e:
return {"status": "error", "detail": str(e)}