Monday, October 12, 2026: Infinigence AI raises $2.3B Series C, cumulative $4.3B, China's largest AI Infra financing. Zhipu GLM-5.3 high-speed version debuts at 490 tokens/s.

Zhipu GLM-5.3 high-speed version is doing 490 tokens/s. Almost 500. Fast inference makes Chinese model APIs cheaper and pushes usage into agent and consumer apps. Huawei AI passes 25 games using RSI, with 56% fewer operations than humans. Physical AI has pulled in $48B in 2026 already. Efficiency gains plus embodied AI capital is where compute demand goes next. Shanghai large model startup secures $1B. Infinigence AI's $2.3B Series C pushes cumulative to $4.3B. Policy risk is not stopping the check writers. China is pairing faster inference with cheap operations and huge infra funding. If that holds, Chinese AI app margins could surprise US comps. Watch Zhipu API pricing and Infinigence compute deals.

Zhipu GLM-5.3 high-speed version is doing 490 tokens/s. Almost 500. Fast inference makes Chinese model APIs cheaper and pushes usage into agent and consumer apps.