OpenAI

OpenAI provides GPT models including GPT-4o for conversational AI.

Setup

1. Get API Key

1. Go to OpenAI Platform
2. Sign in or create an account
3. Navigate to API keys
4. Create a new secret key

2. Configure Environment

LLM_PROVIDER=openai
OPENAI_API_KEY=sk-...your-api-key
OPENAI_MODEL=gpt-4o

Available Models

Model	Context	Best For	Cost (Input)
gpt-4o	128K	Best quality	$2.50/1M
gpt-4o-mini	128K	Cost effective	$0.15/1M
gpt-4-turbo	128K	Previous gen	$10.00/1M
o1	200K	Reasoning	$15.00/1M

Configuration Options

# Required
OPENAI_API_KEY=sk-...

# Model selection
OPENAI_MODEL=gpt-4o

# Optional
OPENAI_ORGANIZATION=org-...     # For org accounts
OPENAI_BASE_URL=https://...     # For proxies/Azure

Using with RAG

OpenAI doesn't have native File Search like Gemini. For RAG:

1. Use Gemini for RAG + OpenAI for generation (hybrid)
2. Use external vector database (Pinecone, etc.)
3. Use OpenAI Assistants API (requires additional setup)

Troubleshooting

"Invalid API key"

• Check your key at OpenAI Platform
• Ensure you have billing set up
• Check for rate limits

"Model not found"

• Ensure you have access to the model
• Some models require specific access (o1, etc.)

Cost Estimation

For a typical 3-minute session with GPT-4o:

Component	Tokens	Cost
System + context	~2500	$0.00625
User messages	~200	$0.0005
Responses	~1200	$0.012
Total	~3900	~$0.02