Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
camelliaPTM
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
2 ms
·
1.
▲
by
camelliaPTM
5mo ago
The prompt went from 44k to 6k tokens, but you're making two extra model calls per round to get there (chunker + working_memory_update). What does the all-in cost comparison actually look like?