Developers are debating the viability of running local large language models for full-time coding work. While popular AI tools remain the standard, some users are reporting successful transitions to local setups for efficiency and privacy. This discussion explores real-world performance metrics, token generation speeds, and the technical hurdles of running powerful open-source models as primary development tools.