Databricks adopts GLM 5.2 as a lower-cost coding engine

Databricks has made GLM 5.2 its default coding engine after benchmarking coding agents on its own multi-million-line codebase. The Decoder reports that the Chinese open-source model matched Anthropic's Opus 4.8 at a lower per-task cost.

The decision illustrates how enterprises are evaluating coding models on internal workloads rather than relying only on public leaderboards. Cost per successful task can matter more than brand or headline benchmark scores.

It also signals growing confidence in open-source and non-U.S. models for daily software engineering work when they perform well in company-specific environments.