Open-Weight AI Coding Model Matches Proprietary Performance at Lower Cost

Databricks' recent benchmarks show that the open-weight GLM 5.2 coding agent performs comparably to Anthropic's proprietary Opus 4.8. Both models achieved an 87% completion rate on a large codebase. Notably, GLM 5.2 offers this performance at a significantly reduced cost per task, making advanced AI coding more accessible.