Reasonable Read Model release from North ML. See the org page: https://huggingface.co/North-ML1 Part of North ML - open weights, honest numbers.