Skip to content

SETUP_LITELLM.md

Latest commit

 

History

History
91 lines (67 loc) · 2.2 KB

SETUP_LITELLM.md

File metadata and controls

91 lines (67 loc) · 2.2 KB
Raw

⚡ LiteLLM Setup for Your Project

✅ What's Already Done

Application code updated to work with LiteLLM instead of OpenAI:

  • ✅ API routes use LITELLM_API_KEY and LITELLM_BASE_URL
  • ✅ Documentation updated
  • ✅ env.example updated

🔧 What You Need to Do

1. Create .env.local File

# In project root, create .env.local file
touch .env.local

2. Add Your Keys to .env.local

Open .env.local and add:

# Supabase (your existing keys)
NEXT_PUBLIC_SUPABASE_URL=your_supabase_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_supabase_anon_key
SUPABASE_SERVICE_ROLE_KEY=your_service_role_key

# LiteLLM (your new keys)
LITELLM_API_KEY=your_litellm_api_key_here
LITELLM_BASE_URL=your_litellm_base_url_here

3. Restart Dev Server

If server is already running, stop it (Ctrl+C) and start again:

npm run dev

✨ Done!

Application will now use Hyperskill's internal LiteLLM proxy instead of OpenAI.

🧪 Testing

  1. Login as admin
  2. Open any saved report
  3. Click "🤖 Generate Manager Report"
  4. If generation successful - everything works! 🎉

🐛 Issues?

Error "Failed to generate report"

  • Check that .env.local contains both parameters:
    • LITELLM_API_KEY
    • LITELLM_BASE_URL
  • Make sure you're on Hyperskill network (VPN may be required)
  • Restart dev server after changing .env.local

Error "Module not found: openai"

npm install

Check Environment Variables

Add temporarily at start of API route for debugging:

console.log('LiteLLM Config:', {
  apiKey: process.env.LITELLM_API_KEY?.substring(0, 10) + '...',
  baseURL: process.env.LITELLM_BASE_URL
});

📊 OpenAI vs LiteLLM Differences

Parameter OpenAI LiteLLM
API Key OPENAI_API_KEY LITELLM_API_KEY
Base URL (default) LITELLM_BASE_URL
Model gpt-4o gpt-4o (same)
API Format OpenAI compatible OpenAI compatible
Cost Direct through OpenAI Through Hyperskill

💡 Notes

  • LiteLLM uses the same API interface as OpenAI
  • The openai npm package is compatible with LiteLLM
  • Only need to specify baseURL when initializing client
  • Model remains gpt-4o as before