@@ -225,11 +225,6 @@ This project uses **Supabase** for authentication and **enterprise-grade securit
225225 OLLAMA_URL=http://localhost:11434
226226 OLLAMA_MODEL=deepseek-r1:7b
227227
228- # Option 2: Use DeepSeek API (recommended for production on Vercel)
229- DEEPSEEK_API_KEY=your-deepseek-api-key
230-
231- # Option 3: Use Gemini API
232- GEMINI_API_KEY=your-gemini-api-key
233228 SMTP_FROM_EMAIL=your-email@gmail.com
234229 SMTP_FROM_NAME=Lab68 Dev Platform
235230
@@ -374,40 +369,6 @@ For production, run Ollama on a separate server:
374369- ✅ Full control - choose any model, customize parameters
375370- ✅ Platform-specific knowledge - AI trained on YOUR docs
376371
377- ---
378-
379- <<<<<<< Updated upstream
380- --- ```bash
381-
382- # Add to Vercel environment variables
383-
384- DEEPSEEK_API_KEY=your-key # Recommended (cheap)
385-
386- # OR
387-
388- GEMINI_API_KEY=your-key # Free tier available
389-
390- ```
391-
392- 2. **Hybrid Setup (Best for High Usage):**
393- - Deploy Next.js on Vercel
394- - Run Ollama on separate VPS (AWS EC2, DigitalOcean, Hetzner)
395- - Point to your VPS:
396-
397- ```env
398- OLLAMA_URL=https://ai.yourdomain.com
399- ```
400-
401- - Users get local model speed + privacy without installing anything
402-
403- ### Smart Fallback System
404-
405- The AI tries providers in this order:
406-
407- ---
408-
409- =======
410- >>>>>>> Stashed changes
411372## Feature Overview
412373
413374| Area | Summary |
0 commit comments