B.AI, a one-stop AGI (artificial general intelligence) infrastructure platform, announced on Aug. 27 that Zhipu's GLM-5.3-Flash (320B-A18B) is officially available on its platform and fully free through the official API group. The first natively multimodal model in the GLM-5 series has 320 billion total parameters and 18 billion activated parameters. It is powered by domestically produced AI chips and introduces a hybrid architecture combining sparse and linear attention for the first time, alongside a 1M-token context window. Developers can use the B.AI official API group at no cost for complex reasoning, long-text processing and multimodal workloads.