ToolsJuly 9, 2026via The Decoder
Databricks makes Chinese open-source model GLM 5.2 its default coding engine after it matched Opus at lower cost
Why it matters
Databricks' internal benchmark reveals open-source models can match frontier commercial models on real-world coding tasks at significantly lower cost, signaling a shift toward pragmatic model selection over brand loyalty. The decision to default to GLM 5.2 demonstrates that companies are moving beyond public benchmarks to build proprietary eval frameworks.
Key signals
- GLM 5.2 matched Anthropic's Opus 4.8 on Databricks' internal coding benchmark
- Cost differential: $1.28 per task (GLM 5.2) vs $1.94 per task (Opus 4.8)
- Benchmark conducted on Databricks' multi-million-line internal codebase
- GLM 5.2 is a Chinese open-source model
- Databricks rolling out GLM 5.2 as daily coding workhorse
- Key strategic insight: no single provider dominates; internal benchmarks outweigh public ones
The hook
$1.28 per task. That's what Databricks paid GLM 5.2 to match Anthropic's Opus—and they're making it their default coding engine.
Databricks benchmarked coding agents on its own multi-million-line codebase and found that the Chinese open-source model GLM 5.2 matched Anthropic's Opus 4.8 at $1.28 per task versus $1.94. The company plans to roll it out as a daily coding workhorse. Its broader takeaway: no single provider dominat…