← Back to context

Comment by frog437

9 hours ago

depending on how large your codebase is, hopefully not. At this point use something like the IX plugin to ingest codebase and track context, rather than from the LLM itself.

1 comment

frog437

Reply
frog437  6 hours ago

This is crazy..

tokensSaved = naiveTokens - actualTokens

  - naiveTokens = 19.4M — what ix estimates it would have cost to answer your queries without graph intelligence (i.e., dumping full files/directories into context)                                    
  - actualTokens = 4.7M — what ix's targeted, graph-aware responses actually used
  - tokensSaved = 14.7M — the difference