For now, the workaround I implemented involves creating four copies of the same report and applying a filter based on the distinct_id.
Specifically, each report filters for IDs ending with a specific character. This effectively shards the data, reducing the volume per report to one-fourth. This works for our current volume, but we will have to revisit this once we scale.
Feature Suggestion: I recommend the Mixpanel team adds pagination support to the API or removes this limit entirely to avoid the need for manual partitioning. I do lots of custom reporting for clients and we keep hitting this limitations.