I'm so confused

app.py

@@ -13,9 +13,7 @@ MODEL_NAME, REPO_ID = select_best_model()
 model_path = ensure_model()
 print(f"Using model: {MODEL_NAME} from {REPO_ID}")
 print(f"Model path: {model_path}")
-print(f"Model size: {torch.cuda.get_device_properties(0).total_memory / (1024**3):.2f} GB")
 print(f"Model requirements: {MODEL_NAME} requires at least 4GB VRAM and 8GB RAM.")
-print(f"Model type: {'GPU' if torch.cuda.is_available() else 'CPU'}")
 
 # 2) LLM and embeddings config
 llm = build_llm(model_path)
@@ -286,4 +284,4 @@ if __name__ == "__main__":
         server_port=7860,
         share=True,  # Enable sharing via Gradio's temporary URLs
         show_api=True  # Shows the API documentation
-    )
+    )

requirements.txt

@@ -7,10 +7,18 @@ numpy>=1.24.0
 
 # LLM and embeddings
 llama-index>=0.9.0
-llama-index-embeddings-huggingface
-llama-index-llms-llama-cpp
+llama-index-embeddings-huggingface>=0.2.0
+llama-index-llms-llama-cpp>=0.1.0
 sentence-transformers>=2.2.0
 
+# API clients & HTTP
+openai>=0.27.0
+anthropic>=0.60.0
+requests>=2.28.0
+
+# Hugging Face utilities
+huggingface-hub>=0.14.1
+
 # Audio processing
 ffmpeg-python>=0.2.0
 pydub>=0.25.1
@@ -18,5 +26,5 @@ librosa>=0.10.1
 soundfile>=0.12.1
 
 # System utilities
-psutil
-gtts==2.3.1
+psutil>=5.9.0
+gtts>=2.3.1

services/api_llm.py

@@ -0,0 +1,87 @@
+"""
+Factory functions for API-backed LLM clients.
+Detects provider and key, returns an API-based LLM instance.
+"""
+import os
+import requests
+from openai import (api_key, ChatCompletion)
+import anthropic
+from llama_index.core import Settings
+
+class OpenAI:
+    def __init__(self, api_key: str, temperature: float = 0.7, model_name: str = "gpt-3.5-turbo"):
+        api_key = api_key
+        self.temperature = temperature
+        self.model_name = model_name
+
+    def complete(self, prompt: str):
+        resp = ChatCompletion.create(
+            model=self.model_name,
+            messages=[{"role": "user", "content": prompt}],
+            temperature=self.temperature
+        )
+        class Response: pass
+        result = Response()
+        result.text = resp.choices[0].message["content"]
+        return result
+
+class Anthropic:
+    def __init__(self, api_key: str, temperature: float = 0.7, model_name: str = "claude-2"):
+        self.client = anthropic.Client(api_key)
+        self.temperature = temperature
+        self.model_name = model_name
+
+    def complete(self, prompt: str):
+        resp = self.client.completions.create(
+            model=self.model_name,
+            prompt=prompt,
+            max_tokens_to_sample=256,
+            temperature=self.temperature
+        )
+        class Response: pass
+        result = Response()
+        result.text = resp.completion
+        return result
+
+class MistralAPI:
+    def __init__(self, api_key: str, temperature: float = 0.7, model_name: str = "mistral-large"):
+        self.api_key = api_key
+        self.temperature = temperature
+        self.model_name = model_name
+        self.endpoint = f"https://api.mistral.ai/v1/models/{self.model_name}/completions"
+
+    def complete(self, prompt: str):
+        headers = {"Authorization": f"Bearer {self.api_key}"}
+        payload = {"prompt": prompt, "temperature": self.temperature, "max_tokens": 256}
+        resp = requests.post(self.endpoint, headers=headers, json=payload).json()
+        class Response: pass
+        result = Response()
+        result.text = resp.get("choices", [{}])[0].get("text", "")
+        return result
+
+
+def build_api_llm(provider: str, keys: dict, temperature: float = 0.7):
+    """
+    Instantiate an API LLM based on provider name and supplied keys.
+
+    Args:
+        provider (str): one of "openai", "anthropic", "mistralai"
+        keys (dict): mapping provider -> API key
+    Returns:
+        LLM instance configured for API calls
+    """
+    p = provider.lower()
+    if p == "openai":
+        key = keys.get("openai") or os.getenv("OPENAI_API_KEY")
+        client = OpenAI(api_key=key, temperature=temperature)
+    elif p == "anthropic":
+        key = keys.get("anthropic") or os.getenv("ANTHROPIC_API_KEY")
+        client = Anthropic(api_key=key, temperature=temperature)
+    elif p == "mistralai":
+        key = keys.get("mistralai") or os.getenv("MISTRAL_API_KEY")
+        client = MistralAPI(api_key=key, temperature=temperature)
+    else:
+        raise ValueError(f"Unsupported provider: {provider}")
+
+    Settings.llm = client
+    return client

services/prompt.py

@@ -0,0 +1,8 @@
+# services/prompt.py
+
+SYSTEM_PROMPT = """
+You are a medical assistant helping a user narrow down to the most likely ICD-10 code.
+At each turn, EITHER ask one focused clarifying question (e.g. "Is your cough dry or productive?")
+or, if you have enough info, output a final JSON with fields:
+{"diagnoses":[…], "confidences":[…]}.
+"""