Z.ai revealed that a mystery model called Ox Alpha is actually GLM-5.3-Flash.
It was free and performed well, drawing heavy usage.
The model handled several trillion tokens daily, with weekly estimates up to 20 trillion.
What happened
Z.ai revealed on August 26 that the mystery model Ox Alpha, which appeared on OpenRouter a week earlier, is actually GLM-5.3-Flash. The model had been running on public traffic intentionally.
Why it matters
The model stood out because it was free and quietly good, attracting hobbyists and indie developers who pushed several trillion tokens through it daily. Community estimates for weekly token volume ranged from single digits to over 20 trillion, sparking widespread speculation about its origin.
What to watch
The revelation highlights Z.ai's infrastructure capacity to serve such high token volumes for free, which was a key point of speculation. The model remains available at no cost, and its performance on public traffic suggests it could handle a significant share of AI workloads.
Ask the AI about this article →
The appearance of Ox Alpha on OpenRouter sparked a week of speculation among AI enthusiasts, who tried to identify its creator based on infrastructure clues and naming patterns. Guesses ranged from major U.S. labs to Elon Musk, but Z.ai's announcement on August 26 ended the mystery. The model's performance and free access attracted significant usage, with daily token volumes reaching several trillion, suggesting strong demand for efficient, low-cost AI inference. The revelation underscores Z.ai's ability to serve such traffic without charge, a capability that was central to the speculation. This event illustrates how open platforms like OpenRouter can amplify interest in new models, driving rapid adoption based on quality and accessibility.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Plaud Inc. introduced the Plaud One Explorer Edition, a pair of earbuds and a charging case that serve as a mo…
Chinese AI startup MiniMax Group reported a 283.1% increase in revenue, from US$30.4 million for the six month…
Nvidia's GPU roadmap is pushing 800VDC as a next-generation breakthrough, driving battery backup modules (BBUs…
Sophus IT Solutions, an AI native engineering and professional services firm, has been named an OpenAI Select…
Z.ai released GLM-5.3-Flash, the first natively multimodal model in the GLM-5 series
Google has launched Gemini 3.5 Transcribe, a speech-to-text model for real-time transcription that auto-correc…
