Personally I use llama3.1:8b or mistral-nemo:latest which have a decent contex window (even if it is less than the commercial ones usually). I am working on a token calculator / division of the content method too but is very early
If you have the compute, might as well use the better model :)
The 3.2 series wasn't the kind of leap that 3.0 -> 3.1 was in terms of intelligence; it was just:
1. Meta releasing multimodal vision models for the first time (11B and 90B), and
2. Meta releasing much smaller models than the 3.1 series (1B and 3B).