I don't see how it would be in OpenAIs selfish interest to release an open source model that sucks. Unless you can cohesively explain how that would work in their favor, it seems a lot smarter to assume that they won't.