Skip to content
AI IntelligenceSep 1, 2026Video
Article

Fireship investigates Ox Alpha, an anonymous model that topped OpenRouter by serving 42 trillion tokens in six days, and reveals...

It as Zhipu's GLM-5.3-Flash. Viewers learn how this model delivers strong performance at roughly 40x lower cost than Claude, and what that means for AI economics and agentic workloads.

Data Cube AI EditorialSource: Fireship
01

Source Brief

Fireship investigates Ox Alpha, an anonymous model that topped OpenRouter by serving 42 trillion tokens in six days, and reveals it as Zhipu's GLM-5.3-Flash. Viewers learn how this model delivers strong performance at roughly 40x lower cost than Claude, and what that means for AI economics and agentic workloads.