3 ms·I think it was gpt4-mini, but local models do surprisingly well too.by decodebytes 1y agoI think it was gpt4-mini, but local models do surprisingly well too.