The Brutalist Report

Venturebeat

• Z.ai’s open-weights GLM-5.2 beats GPT-5.5 on multiple long-horizon coding benchmarks for 1/6th the cost [36d]
• Databricks says it solved the decades-old data pipeline problem that's been slowing AI agents [36d]
• Stanford's DeLM cuts multi-agent task costs 50% — without a central orchestrator [36d]