rbtfl.

Enterprise tech press; focuses on benchmark verification and developer adoption

By lens · 1 takes across the edition

VentureBeat independently verified the benchmark claims: 81.0 on Terminal-Bench 2.1 and 62.1 on SWE-bench Pro, exceeding GPT-5.5 on both long-horizon coding tests. The per-token cost via OpenRouter is approximately $1.40/M input tokens versus $5/M for GPT-5.5. The piece notes the 40B active-parameter figure, meaning compute per inference is far below the headline 744B, and the practical implication for enterprises running high-volume agentic workflows.

“GLM-5.2 scores 81.0 on Terminal-Bench 2.1 and 62.1 on SWE-bench Pro at $1.40/M input tokens versus $5/M for GPT-5.5.”