Hey royinblr11, 

Where did this question come from and when was it published? You are correct in that the latest DBRX model has a 32k token context window, larger than MPT-30B's 8k context window. Our latest publication on this stat was March 2024. If the question was published before then, it might be out of date.

https://www.databricks.com/blog/introducing-dbrx-new-state-art-open-llm