Zhipu AI Releases GLM-5-Turbo: OpenClaw-Scenario Optimized High-Speed Model with 2-3x Speed Boost and 20% API Price Hike

Zhipu AI recently released GLM-5-Turbo, a high-speed optimized variant of its GLM-5 flagship model, deeply customized for OpenClaw agent scenarios. The model retains the 744B MoE architecture, with a 2-3x speed boost over GLM-5, supporting 200K context and 128K maximum output.

Positioned as an "OpenClaw native model", GLM-5-Turbo has been specially optimized in four areas: tool call stability, complex instruction decomposition, time dimension understanding for scheduled/continuous tasks, and high-throughput long-chain execution efficiency. In ZClawBench evaluations, its performance in OpenClaw scenarios is significantly better than GLM-5, leading mainstream models in multiple key tasks.

In terms of commercialization, the API price of GLM-5-Turbo has been increased by 20% simultaneously. Personal "Lobster Packages" are launched: an experience monthly card for 39 yuan (35 million Tokens) and an advanced monthly card for 99 yuan (100 million Tokens). The usage limit of GLM-5-Turbo in the GLM Coding Plan has been tripled, and users can enjoy the same high-capacity quota as GLM-4.7 during off-peak hours. The promotion ends on April 30.