Update app.py

app.py

@@ -1,9 +1,9 @@
 import gradio as gr
 import os
-from huggingface_hub import InferenceClient
+import requests
 
 HF_TOKEN = os.environ.get("HF_TOKEN", "")
-MODEL = "mistralai/Mistral-7B-Instruct-v0.3"
+API_URL = "https://api-inference.huggingface.co/models/mistralai/Mistral-7B-Instruct-v0.3"
 
 SYSTEM_PROMPT = """You are an expert API architect and indie hacker advisor.
 
@@ -20,42 +20,38 @@ SEARCH: Algolia, Typesense, Meilisearch
 AUTH: Auth0, Clerk, Supabase Auth, Firebase Auth
 DATABASE: Supabase, Firebase, PlanetScale, Neon, Upstash
 SCRAPING: Apify, ScraperAPI, Browserless, Firecrawl
-ANALYTICS: Mixpanel, PostHog, Amplitude, Plausible
+ANALYTICS: Mixpanel, PostHog, Amplitude, Plausible"""
 
-When a user describes their product idea, recommend the perfect API stack.
 
-Your response MUST follow this exact format:
+def query_model(prompt):
+    headers = {"Authorization": f"Bearer {HF_TOKEN}"}
 
-## 🎯 Your Idea in One Line
-(restate the idea clearly)
+    full_prompt = f"<s>[INST] {SYSTEM_PROMPT}\n\n{prompt} [/INST]"
 
-## 🔧 Recommended API Stack
+    payload = {
+        "inputs": full_prompt,
+        "parameters": {
+            "max_new_tokens": 1200,
+            "temperature": 0.7,
+            "return_full_text": False
+        }
+    }
 
-For each API (recommend 3-5 total):
+    response = requests.post(API_URL, headers=headers, json=payload, timeout=60)
 
-### [API Name] — [Role in the product]
-- **What it does for you:** (one sentence)
-- **Pricing:** (free tier + paid, be specific with numbers)
-- **Docs:** (exact URL)
-- **Why not alternatives:** (one sentence)
-- **Integration difficulty:** Easy / Medium / Hard
+    if response.status_code == 503:
+        return None, "⏳ Model is loading. Please wait 20 seconds and try again."
+    if response.status_code == 401:
+        return None, "❌ Invalid HF_TOKEN. Check Space Settings → Secrets."
+    if response.status_code == 429:
+        return None, "❌ Rate limit hit. Wait 30 seconds and try again."
+    if response.status_code != 200:
+        return None, f"❌ Error {response.status_code}: {response.text[:200]}"
 
-## 💰 Monthly Cost Estimate
-| Users | Est. Cost |
-|-------|-----------|
-| 100   | $X/month  |
-| 1,000 | $X/month  |
-| 10,000| $X/month  |
-
-## ⚡ Build Order
-1. First: [API] — because...
-2. Second: [API] — because...
-3. Third: [API] — because...
-
-## 🚨 Top Mistake to Avoid
-(the most common mistake with this stack)
-
-Be specific, honest about pricing, always mention free tiers."""
+    result = response.json()
+    if isinstance(result, list) and len(result) > 0:
+        return result[0].get("generated_text", ""), None
+    return None, f"❌ Unexpected response: {str(result)[:200]}"
 
 
 def get_stack(idea, budget, level, market):
@@ -71,39 +67,42 @@ def get_stack(idea, budget, level, market):
             "Free token — no credit card needed!"
         )
 
-    try:
-        client = InferenceClient(token=HF_TOKEN)
-
-        prompt = f"""Product idea: {idea}
-
+    prompt = f"""Product idea: {idea}
 Monthly API budget: {budget}
 Technical level: {level}
 Target market: {market}
 
-Recommend the best API stack for this product."""
+Recommend the best API stack. Use this format:
+
+## 🎯 Your Idea in One Line
+(restate clearly)
 
-        messages = [
-            {"role": "system", "content": SYSTEM_PROMPT},
-            {"role": "user", "content": prompt}
-        ]
+## 🔧 Recommended API Stack
+For each API (3-5 total):
+### [API Name] — [Role]
+- What it does for you: (one sentence)
+- Pricing: (free tier + paid with numbers)
+- Docs: (exact URL)
+- Integration difficulty: Easy / Medium / Hard
 
-        response = client.chat.completions.create(
-            model=MODEL,
-            messages=messages,
-            max_tokens=1500,
-            temperature=0.7
-        )
-        return response.choices[0].message.content
+## 💰 Monthly Cost Estimate
+| Users | Est. Cost |
+|-------|-----------|
+| 100 | $X/month |
+| 1,000 | $X/month |
+| 10,000 | $X/month |
 
-    except Exception as e:
-        err = str(e)
-        if "401" in err or "auth" in err.lower():
-            return "❌ Invalid token. Check your HF_TOKEN in Space Settings."
-        if "429" in err or "rate" in err.lower():
-            return "❌ Rate limit hit. Wait 30 seconds and try again."
-        if "loading" in err.lower():
-            return "⏳ Model is loading (cold start). Wait 20 seconds and try again."
-        return f"❌ Error: {err}"
+## ⚡ Build Order
+1. First: [API] — because...
+2. Second: [API] — because...
+
+## 🚨 Top Mistake to Avoid
+(most common mistake)"""
+
+    result, error = query_model(prompt)
+    if error:
+        return error
+    return result
 
 
 def ask_followup(question, prev):
@@ -112,29 +111,17 @@ def ask_followup(question, prev):
     if not prev or prev.startswith("⚠️") or prev.startswith("❌"):
         return "⚠️ Please generate a stack recommendation first."
 
-    try:
-        client = InferenceClient(token=HF_TOKEN)
-        messages = [
-            {"role": "system", "content": SYSTEM_PROMPT},
-            {
-                "role": "user",
-                "content": (
-                    f"Previous recommendation:\n{prev}\n\n"
-                    f"Follow-up question: {question}\n\n"
-                    f"Answer concisely and helpfully."
-                )
-            }
-        ]
-        response = client.chat.completions.create(
-            model=MODEL,
-            messages=messages,
-            max_tokens=800,
-            temperature=0.7
-        )
-        return response.choices[0].message.content
+    prompt = f"""Previous recommendation:
+{prev[:500]}
+
+Follow-up question: {question}
+
+Answer concisely and helpfully."""
 
-    except Exception as e:
-        return f"❌ Error: {str(e)}"
+    result, error = query_model(prompt)
+    if error:
+        return error
+    return result
 
 
 EXAMPLES = [
@@ -155,7 +142,7 @@ with gr.Blocks(css=css, title="API Stack Finder") as demo:
 # 🔧 API Stack Finder
 **Describe your product idea → Get the perfect API stack, pricing, and build order**
 
-Powered by **Mistral-7B-Instruct** via HuggingFace Inference API — 100% Free.
+Powered by **Mistral-7B** via HuggingFace Inference API — 100% Free.
 """)
 
     with gr.Row():
@@ -168,13 +155,8 @@ Powered by **Mistral-7B-Instruct** via HuggingFace Inference API — 100% Free.
         with gr.Column(scale=1):
             budget_input = gr.Dropdown(
                 choices=[
-                    "$0 (free only)",
-                    "$10/month",
-                    "$30/month",
-                    "$50/month",
-                    "$100/month",
-                    "$500/month",
-                    "Unlimited"
+                    "$0 (free only)", "$10/month", "$30/month",
+                    "$50/month", "$100/month", "$500/month", "Unlimited"
                 ],
                 value="$30/month",
                 label="Monthly API budget"
@@ -207,7 +189,7 @@ Powered by **Mistral-7B-Instruct** via HuggingFace Inference API — 100% Free.
     with gr.Row():
         followup_input = gr.Textbox(
             label="Follow-up question",
-            placeholder="e.g. Is there a free alternative to Stripe? / How do I add auth cheaply?",
+            placeholder="e.g. Is there a free alternative to Stripe?",
             scale=4
         )
         followup_btn = gr.Button("Ask", scale=1, variant="secondary")
@@ -227,8 +209,8 @@ Powered by **Mistral-7B-Instruct** via HuggingFace Inference API — 100% Free.
 
     gr.Markdown("""
 ---
-🤖 Model: `Mistral-7B-Instruct-v0.3` · 🆓 Free via HuggingFace Inference API  
-🔑 Get your free token: [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens)
+🤖 Model: `Mistral-7B-Instruct-v0.3` · 🆓 Free via HuggingFace  
+🔑 Token: [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens)
 """)
 
 if __name__ == "__main__":