Free LLM tokens - one native ad per response
Book flights worldwide. 20% off with code FLY20 at alphaairlines.com.
Studio-quality headphones, 20% off today. techgearpro.com.
Learn to code fast - first lesson free. codemaster.io.
Luxury beachfront stays from $99/night. betaresort.com.
Daily vitamins with free shipping on first order. vitaboost.com.
Meal kits delivered - first box free. freshbite.com.
Implements the generative auction from Zhao et al., "LLM-Auction: Generative Auction towards LLM-Native Advertising" (2025). The full candidate ad set with bids is passed to llama-3.1-8b via Groq. The model selects the most relevant ad and integrates it natively into its response in a single inference call. A first-price payment rule applies. The green debug line shows the auction outcome.
The API is OpenAI-compatible. Just point your client at /api/chat with standard messages:
curl -X POST https://your-host/api/chat \
-H "Content-Type: application/json" \
-d '{"messages": [{"role": "user", "content": "Plan a trip to Hawaii"}]}'
Response includes the LLM reply with a natively integrated ad, plus an auction object showing which advertiser won and their payment. Works as a drop-in replacement for any OpenAI-style chat endpoint - free tokens, ad-supported.