Dev burns $2k on OpenAI, uncovers one feature hogging 70% of the bill—spawns SDK wrapper to tag the LLM money pits

A developer recently spent $2,147 on OpenAI's GPT-4 model before discovering that one feature accounted for approximately 70% of the total cost. To better track expenses, the developer created a tool that tags Large Language Model (LLM) calls with feature names, enabling more granular cost monitoring. The solution, called Orbit, is an SDK that wraps existing clients without requiring code changes to API calls. It supports integration with OpenAI, Anthropic, and Gemini, and offers a free tier with no credit card required. The Orbit SDK can be installed via npm using the @with-orbit/sdk package, and documentation is available on the company's website. This development highlights the importance of cost transparency and management in the rapidly growing field of artificial intelligence and machine learning, where unexpected expenses can quickly add up. By providing a simple and free solution, Orbit aims to help developers optimize their LLM usage and reduce costs.