A new artificial intelligence model has caused a wave of confusion among experts, as it is available for free and has achieved performance exceeding expectations, while the identity of its developer remains unknown, raising questions about its goals and sources of funding.
“Ox Alpha” appeared last week on the OpenRouter AI platform, bearing a label indicating that it is a “steal model” and that it was “developed and operated by an external party that chose not to disclose its identity during this preview.”
The same website explains that the queries sent by users to the system are "retained by the provider," but confirms that they are "not used for training," making it difficult to know the fate of any data entered into the system.
Despite the ambiguity, developers have already begun using it, as the system can process text, images, and videos, and is available for free with almost unlimited use, despite its relatively large power.
Technology experts have praised the model, which has shown performance on par with some of the most well-known models from major companies such as Anthropic and OpenAI. Patrick Collison, CEO of Stripe, described it as "quite amazing," while analysts rushed to try to find out who built it and what its emergence might mean in the global race to build the most powerful artificial intelligence tools.
The emergence of the mysterious system led to widespread speculation among experts and commentators about its creators. Initial online speculation focused on Chinese companies, despite weak evidence. In the past, Chinese AI companies have launched their systems with little or no prior announcement, and some have suggested that the Ox Alpha model might have been developed by a Chinese company like Z.ai, which previously released another model called GLM-5 under a different name. Other Chinese companies, from ByteDance (the developer of TikTok) to Alibaba (often described as the Amazon of China), have also released their models without immediately claiming responsibility.
