Put that way it's basically competitive evolutionary pressure to exhaust the context window of the other LLM.