Re:Microsoft starts canceling Claude Code licenses
by brookst on Friday May 22, 2026 @03:42PM
(#48240556)
I get 98.6% cache hits on Claude code. Short of drastic arch changes it’s hard to imagine it getting much better.
[ Parent
| on HN ]
Re:Microsoft starts canceling Claude Code licenses
by gobdovan on Friday May 22, 2026 @03:57PM
(#48240773)
98.6% cache hits doesn't distinguish an efficient workflow from an overly chatty linear agent repeatedly reusing the same context. Plus, it says nothing directly that the process has good useful progress per token.
Re:Microsoft starts canceling Claude Code licenses
by kridsdale1 on Friday May 22, 2026 @04:14PM
(#48241022)
We are all going to be graded by (tickets closed / tokens burned) soon enough.
Re:Microsoft starts canceling Claude Code licenses
by recursive on Friday May 22, 2026 @04:55PM
(#48241531)
Sweet. I can get that up to infinity, assuming they're using IEEE-754 division.
Re:Microsoft starts canceling Claude Code licenses
by nchie on Saturday May 23, 2026 @04:24AM
(#48245860)
I doubt it, the difference between someone slightly inefficient and someone extremely efficient isn't big enough to matter compared to how much they cost in salary.
Re:Microsoft starts canceling Claude Code licenses
by hedgehog on Friday May 22, 2026 @05:20PM
(#48241822)
You pay for cache hits on every turn and even with the newest architectures longer context is slower/more energy intensive. Constructing concise turns that reuse prefix and stop when the new context is no longer useful help, as does pushing generation down into cheaper models while using stronger models for verification.