Nobody would trust one of the labs that have their own models with this. They’d assume they’re taking all the data, and biasing the routing to favour themselves. xAI would probably be the least trusted, but the same would apply for any of them.