4 ms·
I have a couple projects where the background research is easily over 500k without writing any code, after ultracode subagents synthesis.
by _zoltan_ 3mo ago
I have a couple projects where the background research is easily over 500k without writing any code, after ultracode subagents synthesis.
- giancarlostoro 3mo agoI think the distinction is when the model decides to read code top to bottom vs when the model chooses to parse code indirectly to save on tokens, there's also AST tooling to let the model see project structure.