Update app.py

app.py

@@ -1,36 +1,39 @@
 from fastapi import FastAPI
 from pydantic import BaseModel
-from lmstudio import LLM  # LM Studio Python interface
-from fastapi.middleware.cors import CORSMiddleware
-
-# FastAPI app
-app = FastAPI(title="ChatGPT OS 1.0", description="Local AI Chat API", version="1.0")
-
-# Enable CORS so frontend can connect from any origin
-app.add_middleware(
-    CORSMiddleware,
-    allow_origins=["*"],
-    allow_methods=["*"],
-    allow_headers=["*"]
+from llama_cpp import Llama
+import os
+
+app = FastAPI()
+
+MODEL_PATH = "./models/gpt-oss-20b-Q3_K_M.gguf"
+
+print("🔄 Loading model… this may take a while")
+
+llm = Llama(
+    model_path=MODEL_PATH,
+    n_ctx=16384,
+    n_threads=os.cpu_count(),
+    n_gpu_layers=0,  # HF CPU-only unless paid GPU
+    verbose=False,
 )
 
-# Load the GPT-OSS model
-model_path = "./models/gpt-oss-20b-Q3_K_M.gguf"
-llm = LLM(model_path=model_path, context_length=16384, flash_attention=True)
+print("✅ Model loaded successfully")
 
-# Request body schema
 class ChatRequest(BaseModel):
-    message: str
+    prompt: str
 
-# Chat endpoint
 @app.post("/chat")
-async def chat(req: ChatRequest):
-    user_message = req.message
-    # Generate response from the model
-    response = llm.generate(user_message, max_new_tokens=256)
-    return {"response": response}
-
-# Optional: health check
-@app.get("/health")
-def health():
-    return {"status": "ok", "model_loaded": True}
+def chat(req: ChatRequest):
+    output = llm(
+        f"User: {req.prompt}\nAssistant:",
+        max_tokens=512,
+        stop=["User:"],
+    )
+
+    return {
+        "response": output["choices"][0]["text"].strip()
+    }
+
+@app.get("/")
+def root():
+    return {"status": "ChatGPT Open-Source 1.0 is running 🚀"}