Exported datasets are being truncated, making them unusable

Gemini API
, , , ,
1

When exporting a dataset, the text content for requests and responses is being trimmed to roughly 205–230 characters, rendering the exported data unusable.

Additional Context

  • The full input/output content is successfully captured by the system (verified by manually checking the logs).

  • The truncation appears to happen specifically during the export generation process.

Expected Behavior

The export utility should output the complete, unclipped text for all request and response fields.

2

Update: Relocated from AI Studio to the Gemini API topic

3

Still no follow-ups on this topic. I will report if there is any new information or a resolution of this