QuestionAI / MLAI / ML. models, prompts, intelligence resolved
how do I not blow my token budget on every request?
I send the whole conversation history each time and its getting expensive. whats the standard way to trim context?
2 answers
I send the whole conversation history each time and its getting expensive. whats the standard way to trim context?
do not overthink it — the simple version is fine for now, optimize once someone hits it.
shipped the same thing last month. the gotcha is the async bit. happy to look at your code.