GLM-5.2 Leads Open-Model Benchmark With 1M Token Context

Z AI’s GLM-5.2 has topped the Artificial Analysis Intelligence Index for open-weight models, scoring 51—the highest open-model score reported. It edges past GLM-5 (50). Key upgrades in GLM-5.2 include a 1 million token context window, up from 200K in GLM-5.1. The company also emphasizes coding and long-term agentic tasks, adding dual “thinking-effort” levels to let developers trade off speed versus depth. Z AI positions GLM-5.2 as a top performer on benchmarks such as FrontierSWE and PostTrainBench, which test real-world software engineering and post-training task completion. The release is the fourth major GLM iteration in roughly four months, following GLM-5 and GLM-5.1’s earlier gains. Access is planned via tiered API and chatbot offerings under the GLM Coding Plan. The article notes that independent evaluations reportedly validate the performance claims, though Z AI had not published official benchmarks at the initial June 13 launch. A notable point for investors: the model release includes no blockchain or cryptocurrency integration, separating it from the current trend of fintech-style crypto features in AI systems.