The tech world has a new mystery on its hands. A previously unknown AI model called Ox Alpha appeared on OpenRouter Thursday, described only as a “stealth model” from an anonymous third-party provider. No one is saying who built it — and that hasn’t stopped developers from putting it to work.
OpenRouter bills Ox Alpha as a reasoning model designed for coding, sustained agentic work, and production workloads. It is built for long-horizon software engineering, complex reasoning, and workflows that combine text with visual context. Perhaps most striking? It’s free.
OpenCode, an open-source AI coding agent, announced on X that Ox Alpha would remain free for a week, with “near unlimited usage” and a provider capacity of 100 trillion tokens per day. That figure is roughly 100 times the number of AI tokens Visa said it processes in an entire month.
The model has already drawn attention from prominent tech figures. Stripe CEO Patrick Collison tried Ox Alpha and posted on X that “it’s very impressive.”
Where does Ox Alpha come from? Early speculation points to China. Wccftech, an online tech publication, suggested the lab behind GLM-5, Z.ai, as a possible origin. That company previously tested GLM-5 anonymously under the name “Pony Alpha,” and developers say Ox Alpha’s tokenizer behavior and responses bear similarities.
The timing would make sense. Chinese AI labs like Zhipu, DeepSeek, and Moonshot AI are increasingly challenging US rivals. Their models have approached the performance of leading American systems at a fraction of the cost, and many are open-source. Moonshot’s Kimi K3, released in July, is a 2.8 trillion-parameter open-weight model built for coding, reasoning, and agentic tasks. It quickly drew attention in Silicon Valley for its performance and lower price.
Still, the evidence is far from conclusive. Wccftech later highlighted a competing analysis suggesting Ox Alpha’s tokenizer could instead point toward Microsoft’s MAI family. On Saturday, prominent AI analyst Andrew Curran wrote on X that GLM was the leading theory on Friday night, but by Saturday morning, “people seem less sure of anything.”
