9 comments

[ 4.3 ms ] story [ 28.3 ms ] thread
Have they published?
It’s probably something like deepseek’s native sparse attention with content based granularity. They might not be publishing anything because it’s not such a strong value proposition and doing so would lead to commentary that would tank their investment opportunities.
For Claude Code, I feel 1M is enough. I've had a compaction once, but that was because I was forcing Claude to do something it clearly had a hard time understanding.

For general chat bots where the user doesn't understand what a context window is, what do you do about context? Latest few messages and then a memory tool? Compaction?

I use a tool called context-mode. Update agent to save session summary every 100k tokens
Feels like this is to AI what JPG is to image
Waiting for the paper and model card. I believe it when I see it.