API reference
This section contains the full API specification for Braintrust's data plane. The API is hosted globally at https://api.braintrust.dev or in your own environment. The API allows you to access all of the core objects in Braintrust, including experiments, datasets, prompts, users, groups, roles, and more. It also enables you to access Braintrust from languages other than Typescript and Python.
You can access the full OpenAPI spec for this API at https://github.com/braintrustdata/braintrust-openapi.
Acls
Permission management for resources
ApiKeys
Management of API keys for Braintrust users
CrossObject
Events and feedback across object types
Datasets
Collections of data used to run evaluations and track improvements over time
Evals
Run evaluations through the API
Experiments
A set of traces run to test the behavior of code
Functions
Invokable custom logic for LLM applications
Groups
Permission management for collections of users
Logs
Log real-world interactions in your application
Organizations
The billing unit of Braintrust. Can represent a company or team
Other
'Hello world' endpoint for testing the API
ProjectScores
Scores scoped to a project to be used in experiment code and human review
ProjectTags
Tags used to track various kinds of data across your application, and track how they change over time
Projects
A single AI feature. The container for experiments, logs, datasets, prompts, and playgrounds
Prompts
Versioned prompt that can be referenced in your code
Proxy
An OpenAI-protocol compatible proxy that supports multiple model formats, caching, and secret management
Roles
Collections of permissions that can be assigned to users or groups
Users
Braintrust users
Views
Table configurations that can be saved for quick access
API wrappers
Through Stainless, we have language-specific wrappers over the bare REST API for a variety of languages. Note that unlike our custom-built Python or Typescript SDKs, these libraries map essentially 1:1 with the REST API: