We use TensorZero as our LLM Gateway and prompt evaluation platform. I would think a model-neutral prompt evaluation platform is still largely needed. Therefore I think we should somehow continue its life as a community project (TensorOne would be a good name).
