Update app.py

app.py

@@ -3,125 +3,124 @@ import os
 import requests
 
 HF_TOKEN = os.environ.get("HF_TOKEN", "")
-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.
-
-You know every major API deeply:
-
-PAYMENT: Stripe, Paddle, LemonSqueezy, PayPal, Razorpay
-AI/ML: OpenAI, Anthropic Claude, Replicate, HuggingFace Inference, AssemblyAI, ElevenLabs
-COMMUNICATION: Twilio (SMS/Voice), SendGrid (email), Resend, Mailgun, WhatsApp Business
-DATA/ENRICHMENT: Clearbit, Apollo.io, Hunter.io, Proxycurl
-MAPS/LOCATION: Google Maps, Mapbox, HERE, OpenStreetMap
-MEDIA: Cloudinary, Mux (video), Imgix, Transloadit
-FINANCE: Alpha Vantage, Polygon.io, CoinGecko, Plaid, Finnhub
-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"""
 
+# Using a reliable free model on HuggingFace
+API_URL = "https://api-inference.huggingface.co/models/HuggingFaceH4/zephyr-7b-beta"
 
 def query_model(prompt):
+    if not HF_TOKEN:
+        return None, (
+            "⚠️ **HF_TOKEN is not set.**\n\n"
+            "Space Settings → Secrets → New secret\n"
+            "- Name: `HF_TOKEN`\n"
+            "- Value: your token from huggingface.co/settings/tokens"
+        )
+
     headers = {"Authorization": f"Bearer {HF_TOKEN}"}
 
-    full_prompt = f"<s>[INST] {SYSTEM_PROMPT}\n\n{prompt} [/INST]"
+    full_prompt = f"<|system|>\nYou are an expert API architect and startup advisor.\n</s>\n<|user|>\n{prompt}\n</s>\n<|assistant|>\n"
 
     payload = {
         "inputs": full_prompt,
         "parameters": {
-            "max_new_tokens": 1200,
+            "max_new_tokens": 1000,
             "temperature": 0.7,
-            "return_full_text": False
+            "do_sample": True,
+            "return_full_text": False,
+            "stop": ["<|user|>", "</s>"]
         }
     }
 
-    response = requests.post(API_URL, headers=headers, json=payload, timeout=60)
+    try:
+        response = requests.post(API_URL, headers=headers, json=payload, timeout=60)
+
+        if response.status_code == 503:
+            return None, "⏳ Model is loading (cold start). Wait 30 seconds and try again."
+        if response.status_code == 401:
+            return None, "❌ Invalid HF_TOKEN. Go to Space Settings and check your token."
+        if response.status_code == 429:
+            return None, "❌ Rate limit. Wait 1 minute and try again."
+        if response.status_code != 200:
+            return None, f"❌ API Error {response.status_code}: {response.text[:300]}"
+
+        data = response.json()
 
-    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]}"
+        if isinstance(data, list) and len(data) > 0:
+            text = data[0].get("generated_text", "").strip()
+            if text:
+                return text, None
 
-    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]}"
+        if isinstance(data, dict) and "error" in data:
+            return None, f"❌ Model error: {data['error']}"
+
+        return None, f"❌ Empty response from model. Try again."
+
+    except requests.exceptions.Timeout:
+        return None, "⏳ Request timed out. The model may be loading. Try again in 30 seconds."
+    except Exception as e:
+        return None, f"❌ Error: {str(e)}"
 
 
 def get_stack(idea, budget, level, market):
     if not idea.strip():
         return "⚠️ Please describe your product idea."
 
-    if not HF_TOKEN:
-        return (
-            "⚠️ **HF_TOKEN is not set.**\n\n"
-            "Go to: **Space Settings → Secrets → New secret**\n"
-            "- Name: `HF_TOKEN`\n"
-            "- Value: your token from [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens)\n\n"
-            "Free token — no credit card needed!"
-        )
+    prompt = f"""I want to build: {idea}
+
+My monthly API budget: {budget}
+My technical level: {level}
+My target market: {market}
 
-    prompt = f"""Product idea: {idea}
-Monthly API budget: {budget}
-Technical level: {level}
-Target market: {market}
+Please recommend the best API stack for my product using this format:
 
-Recommend the best API stack. Use this format:
+## 🎯 My Idea
+(one sentence summary)
 
-## 🎯 Your Idea in One Line
-(restate clearly)
+## 🔧 Recommended APIs
 
-## 🔧 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
+### [API Name] — [what it does in my product]
+- What it does: (one sentence)
+- Pricing: (free tier details + paid pricing)
+- Docs: (URL)
+- Difficulty: Easy / Medium / Hard
+
+(repeat for 3-5 APIs)
 
 ## 💰 Monthly Cost Estimate
-| Users | Est. Cost |
-|-------|-----------|
-| 100 | $X/month |
-| 1,000 | $X/month |
-| 10,000 | $X/month |
+- 100 users: $X/month
+- 1,000 users: $X/month
+- 10,000 users: $X/month
 
 ## ⚡ Build Order
-1. First: [API] — because...
-2. Second: [API] — because...
+1. Start with [API] because...
+2. Then add [API] because...
+3. Finally [API] because...
 
-## 🚨 Top Mistake to Avoid
-(most common mistake)"""
+## 🚨 Biggest Mistake to Avoid
+(one paragraph)"""
 
     result, error = query_model(prompt)
     if error:
         return error
-    return result
+    return result if result else "❌ No response received. Please try again."
 
 
 def ask_followup(question, prev):
     if not question.strip():
         return ""
     if not prev or prev.startswith("⚠️") or prev.startswith("❌"):
-        return "⚠️ Please generate a stack recommendation first."
-
-    prompt = f"""Previous recommendation:
-{prev[:500]}
+        return "⚠️ Generate a stack recommendation first, then ask follow-up questions."
 
-Follow-up question: {question}
+    prompt = f"""Based on this API stack recommendation:
+{prev[:600]}
 
-Answer concisely and helpfully."""
+Answer this follow-up question clearly and concisely:
+{question}"""
 
     result, error = query_model(prompt)
     if error:
         return error
-    return result
+    return result if result else "❌ No response. Please try again."
 
 
 EXAMPLES = [
@@ -142,14 +141,14 @@ 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** via HuggingFace Inference API — 100% Free.
+Powered by **Zephyr-7B** via HuggingFace Inference API — 100% Free.
 """)
 
     with gr.Row():
         with gr.Column(scale=3):
             idea_input = gr.Textbox(
                 label="Describe your product idea",
-                placeholder="e.g. A platform where freelancers can sell services and get paid instantly...",
+                placeholder="e.g. A platform where freelancers sell services and get paid instantly...",
                 lines=4
             )
         with gr.Column(scale=1):
@@ -209,8 +208,8 @@ Powered by **Mistral-7B** via HuggingFace Inference API — 100% Free.
 
     gr.Markdown("""
 ---
-🤖 Model: `Mistral-7B-Instruct-v0.3` · 🆓 Free via HuggingFace  
-🔑 Token: [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens)
+🤖 Model: `Zephyr-7B-Beta` · 🆓 Free via HuggingFace  
+🔑 Get token: [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens)
 """)
 
 if __name__ == "__main__":