# What is Braintrust? Definition and meaning

A platform for running experiments, scoring outputs and tracking quality over time.

## Definition

**Braintrust**: A platform for running experiments, scoring outputs and tracking quality over time.

In short: Evaluation and observability platform for AI products.

## When it is used

Regression-testing prompts and models before release.

## Where it fits

Braintrust belongs to **LLM observability and evaluation** in the [Generative AI and LLMs](/topics/generative-ai-llms) topic. These are the other terms in the same group:

- **[LangSmith](/glossary/langsmith)**: A platform from LangChain for tracing runs, building datasets and evaluating prompts and agents.
- **[Langfuse](/glossary/langfuse)**: A platform for tracing, prompt management, scoring and cost tracking of LLM applications.
- **[Phoenix](/glossary/phoenix)**: A tool for tracing LLM applications with OpenTelemetry and running evaluations on the traces.
- **[PromptFoo](/glossary/promptfoo)**: A command-line tool that runs test cases and comparisons across prompts and models, including security tests.
- **[lm-evaluation-harness](/glossary/lm-evaluation-harness)**: EleutherAI's framework that runs a large library of academic benchmarks against a model with a unified interface.

## Learn more

The full guide to **[Braintrust](/tools/braintrust)** explains how it works, when to use it and how it compares with the alternatives, with examples. The [glossary](/glossary) lists every term on the site from A to Z.

Previous term: [BrainScaleS](/glossary/brainscales) · Next term: [Bronze layer](/glossary/bronze-layer)
