Comment by chuckcode

Would like to see the latency and cost of parsing entire 10M context before throwing out the RAG stack which is relatively cheap and fast.