Most Chinese models are inherently more secure than Anthropic or OpenAI's offerings. Not because they're Chinese, but because they're open-weight. If you value

Most Chinese models are inherently more secure than Anthropic or OpenAI's offerings. Not because they're Chinese, but because they're open-weight. If you value data security and ensuring you're not sending data off to servers, then Anthropic and OpenAI are off the table. The security they offer is just far too weak; you have to rely on blind trust. If you deploy a self-hosted Chinese model, then you know exactly where the data is going. "but nobody is self-hosting Chinese models!" - yes they are, at scale, in datacenters within the US. Many companies already own and operate datacenters; it's a no-brainer.