"The emergence of 'Niu Lai' has ignited a global developer frenzy, with Google employees stepping in to claim their share, while the identity of the domestic model remains to be unveiled."

The anonymous model Ox Alpha (referred to as "Niu Lai" in the Chinese community) has seen a rapid increase in usage since its recent launch on OpenRouter, quickly rising to the top of the platform and breaking the record for single-day model usage. According to the latest data from OpenRouter, the aggregate calls from just the top five applications, including Hermes Agent and Claude Code, have surpassed 4 trillion tokens, with its popularity rapidly spreading from community enthusiasts to global developers and real-world workflows for agents. It is reported that this model supports around 1 million tokens for context and can process text, images, and video inputs, garnering impressive recognition overseas. Patrick Collison, CEO of Stripe, posted his evaluation stating it is "very impressive." The latest complete score of 63 in the 113-item DeepSWE test indicates its overall performance is close to that of Claude Opus 4.8. The identity of the manufacturer of Ox Alpha remains unclear, with technical fingerprints pointing more towards the Z.AIGLM series. Meanwhile, several researchers from Google DeepMind have hinted in posts that it may be a new version of Gemini, but no official claim from any manufacturer has been made as of yet. Notably, OpenCode has stated that it has prepared a service capacity of 100 trillion tokens per day for Ox Alpha. If it is confirmed to come from a Chinese model manufacturer, it will further validate the capability of domestic large models to directly enter the global developer ecosystem and provide global distribution and inference. This could enhance market attention towards domestic models, AI inference infrastructure, and the agent application industry chain. It is reported that the free window for Ox Alpha is expected to end around August 26.