The GLM-5 AI model, developed by Z.ai, has established itself as the most advanced open weights model available. As reported by Artificial Analysis, GLM-5 operates at a level comparable to Claude Opus 4.5 while being priced approximately 5 to 7 times lower. It also surpasses competitors such as Gemini 3 Pro and GPT-5.2 Codex in performance. Among its notable features, GLM-5 exhibits an impressive accuracy rate, achieving a score of 66% on the AA-Omniscience benchmark, which assesses the likelihood of AI generating false information. In this category, Claude Sonnet 4.5 ranks second with a score of 52%. The model excels in coding tasks, performing on par with leading closed-source models, and even achieving superior results in certain scenarios. However, when it comes to scientific applications, it still lags behind GPT and Claude models. Despite its open weights nature, running GLM-5 on personal hardware is not feasible, as it comprises 744 billion parameters (with 40 billion active), necessitating hundreds of gigabytes of VRAM. Nevertheless, users can access GLM-5 for free on the Z.ai platform.
Informational material. 18+.
