Ask HN: How are you managing context loss when switching LLMs to save tokens?

I'm hitting a wall with model context management. On free or lower-tier API plans, those token limits vanish incredibly fast when you're passing back-and-forth conversation history.