The newly announced Jupyterlite notebook widget lets you embed notebooks (very similar to what colab provides) in a Grist document where they can directly access data without having to use CSVs or anything. You can use the Seaborn library as well as many other Python plotting libraries, although I’ve found the Seaborn library specifically to not work nicely inside a @grist.on_records callback so it’s not great if you want the chart to automatically update when data changes. But matplotlib, plotly, altair, and probably others work well, or you can just use Seaborn and rerun the cell manually.