What are some budget-friendly options for adding AI search to a side project?

The price range you've found is pretty standard for general LLM APIs. However, if you look into dedicated search APIs like Tavily, you might find them as low as $0.80 per 1,000 queries. That's about 10 times cheaper since they handle the web grounding separately from generating responses. If your project gets fewer than 5,000 queries a month, you could also take advantage of the free tier from Vertex AI Gemini, although you'll sacrifice some answer quality compared to a more robust search-then-synthesize approach. Also, have you thought about caching results? Implementing a simple Redis cache for frequently repeated queries could cut your API costs by 60-70%!

What exactly do you need the AI for? The better you can define that, the easier it might be to find a cheaper solution tailored to your needs!

Definitely check out some smaller or newer AI startups. They often offer competitive pricing and may have trial periods or lower quotas. Plus, they might not have the same overhead as big players in the market.