3 ms·
Question for those using local models for coding assistance: how well do the best locally runnable models (running on a laptop with a GPU) work for the easy ca
by danans 1y ago
Question for those using local models for coding assistance: how well do the best locally runnable models (running on a laptop with a GPU) work for the easy case:
Writing short runs of code and tests after I give an clear description of the expected behavior (because I have done the homework). I want to save the keystrokes and the mental energy spent on bookkeeping code, not have it think about the big problem for me.
Think short algorithms/transformations/script, and "smart" auto complete.
No writing entire systems/features or creating heavily interpolated things due to underspecified prompts - I'm not interested in those.
- tomck 1y agoI have tried a model on my laptop+GPU before, and it is incredibly unusable. Incredibly slow and just bad output for exactly the work you describe If you're looking for a cheap practical tool + don't care if it's not local, deepseek's non-reasoning model via openrouter is the most cost efficient by far for the work you describe. I put 10 dollars in my account about 6 months ago and still haven't gotten through it, after heavy use semi regularly.