LLM prices dropped 80% — but are you actually saving money?

veryone is cheering about Anthropic and OpenAI dropping API prices by 80%. It sounds great on Twitter. But if you look at your actual billing dashboard, your costs probably haven't moved that much. Why? Because cheaper tokens usually just mean you start wasting more tokens. Here is the thing: 1- Context bloat When GPT-4 was expensive, we carefully truncated histories and compressed prompts. Now that it's cheap, devs just throw the entire 128k context window at it on every single retry. The co...