Dataset search limited to 1000 records

Applies to:

Plan - Any
Deployment - Braintrust-hosted, Self-hosted
Use case - Projects with more than 1000 datasets when searching from the Datasets list page

Summary

Issue: Dataset search in the Datasets list page only returns results from the most recent 1000 records, making older datasets unsearchable in the UI. Datasets created earlier than the 1000th most recent record do not appear in search results. Cause: The client-side in-memory database limits record loading to 1000 entries for performance reasons. Resolution: Use the experiment creation flow to search all datasets, or use the Braintrust API for programmatic access.

Resolution Steps

Workaround 1: Use the Braintrust API

Step 1: Query datasets via API

Use the Braintrust API to search and filter datasets programmatically without UI limitations.

import braintrust

client = braintrust.Braintrust()
datasets = client.datasets.list(project_id="your-project-id")

# Filter for specific dataset
target_dataset = [d for d in datasets if "dataset-0000" in d.name]

Workaround 2: Use experiment creation flow

Step 1: Navigate to experiment creation

Create a new experiment in your project and use the “Add to dataset” search functionality, which searches across all datasets without the 1000-record limit.

Step 2: Search for your dataset

Use the dataset search in the experiment creation flow to locate and select your dataset.

​Summary

​Resolution Steps

​Workaround 1: Use the Braintrust API

​Step 1: Query datasets via API

​Workaround 2: Use experiment creation flow

​Step 1: Navigate to experiment creation

​

​Step 2: Search for your dataset

​

Summary

Resolution Steps

Workaround 1: Use the Braintrust API

Step 1: Query datasets via API

Workaround 2: Use experiment creation flow

Step 1: Navigate to experiment creation

Step 2: Search for your dataset