Corporate America is getting hooked on open-source AI
Points and comments are a snapshot, not live.
Article body wasn't reachable. The HN discussion summary is below.
Points and comments are a snapshot, not live.
Article body wasn't reachable. The HN discussion summary is below.
What commenters are saying
Commenters split on whether open models are sufficient for corporate coding work. Proponents argue cheap open models (e.g., self-hosted Qwen 3.8) excel at common CRUD tasks, costing ~$100/month versus $45k/year for SOTA tokens, fitting roles in the 60-85k compensation range. Critics note SOTA models like Opus 4.5 are needed for competitive software development, though some cite Kimi K3 and GLM5.3 beating Opus 4.5. A Linux kernel developer reported open models already find kernel bugs with acceptable false positive rates. Other discussions touched on copyright risks, legal certainty for US firms avoiding Chinese models, and political fealty factors under the current U.S. regime.
A separate thread argued IP abolitionism conflicts with open-source licenses' reliance on copyright, while another noted Chinese open-source libraries sometimes use license-washing with improperly licensed models.