Feels like this is to AI what JPG is to image
The context window has been shattered: Subquadratic debuts a 12M token window
11–15 of 15 posts
Re: The context window has been shattered: Subquadratic debuts a 12M token window
#12For 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?
This is so interesting to me, I frequently experience compaction on long running features and still find Claude is better than starting with fully fresh prompts.
Every dev seems to use these tools differently.
Re: The context window has been shattered: Subquadratic debuts a 12M token window
#13Waiting for the paper and model card. I believe it when I see it.
Re: The context window has been shattered: Subquadratic debuts a 12M token window
#14For 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 feel the 1M context is way too large —- the model gets ”drunk” way before it gets anywhere near 1 million. Imo the 1M context window is a huge downgrade.
Re: The context window has been shattered: Subquadratic debuts a 12M token window
#15For 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