This announcement does not provide any benchmarks, so it is impossible to tell how useful the model is.