> I believe it will get better and more efficient as we go. Yes of course. The point remains: the LLM has to process the data somehow. If you are concerned ab

> I believe it will get better and more efficient as we go. Yes of course. The point remains: the LLM has to process the data somehow. If you are concerned about costs and token usage then switch to a provider that works for your problem (Flash Gemini looks very interesting..)