Skip to index

Braintrust

Product status: Status not recorded

No lifecycle review is recorded for this listing. Its inclusion in the directory does not confirm current availability; check the vendor before relying on it.

No status check date recorded

AI evaluation platform for datasets, experiments and scorers, with hosted service and enterprise self-hosted data-plane options. No lifecycle review is recorded for this listing. Its inclusion in the directory does not confirm current availability; check the vendor before relying on it.

Product preview

www.braintrust.dev
Preview of Braintrust

Decision notes

Official-document research

Based on the cited documentation. We did not use the product, purchase a plan, or run quality, speed or reliability tests in this review.

Consider it for

  • Teams maintaining repeatable AI evaluation datasets

Choose another approach if

  • Fully offline operation or quality guarantees without curated examples

Pricing and conditions

Documented terms

Starter $0/month: $10 model credits, 1GB processed then $4/GB, 10,000 scores then $2.50/1,000, retention 14 days. Pro $249/month: $100 credits, 5GB then $3/GB, 50,000 scores then $1.50/1,000, retention 30 days then $0.50/GB/month up to 180 days. Enterprise quoted. Applicable paid terms specify USD, tax extra; regional totals and order billing remain unconfirmed. Model/Topics credits differ from score counts; own-provider keys have external fees. MSA defaults to annual advance invoicing unless the order differs; monthly list prices do not guarantee installment billing. Startup offers have separate eligibility.

Pricing source read:

Pricing or terms source

Code, model and service licensing

Apache-2.0 covers the SDK with notice obligations, not the proprietary hosted platform. The MSA preserves customer data ownership, permits service processing, and expressly excludes AI/ML training, development or improvement on Customer Data. It still permits nonidentifying Usage Data for improvement and retention under configuration/legal exceptions. Do not extend the MSA promise to every free account: the July 2026 general ToS permits service-operation content use without repeating that express no-training clause. Executed contracts take priority. External models, data and outputs retain separate rights and processing terms. The July 2026 ToS §3.1 limits Individual Users to personal non-commercial use and Business Customers to internal business use; a separately executed agreement governs conflicts.

Basis and limits of the recorded score

The existing 4.6 editorial score is retained without its original scoring worksheet. This document review did not re-score or measure performance.

Overview

Braintrust organizes evaluations around data, a task and scorers. Code checks or model judges produce experiment scores, with SDK/CI integration for comparing changes. Dataset coverage and scorer design determine what results mean; evaluation support is not proof of production quality or that every regression will be detected.

Hosted plans meter processed data, scores, retention and model use separately. Enterprise self-hosting covers the data plane; UI, authentication and organizational metadata remain in Braintrust's SaaS control plane. Provider requests can leave the VPC, and organization-level provider secrets can reside in that control plane. Review the architecture and applicable contract before logging confidential inputs or outputs.

Sources

The factual claims in this review were read from the pages below. Each date is when an editor last opened the source, not when the product changed.

Pros / Cons

Pros

  • Documented experiment/scorer workflow
  • Enterprise data-plane deployment choice

Cons

  • Requires evaluation-data maintenance
  • Self-hosting retains a SaaS control plane

Who it's for

01

Prompt-change experiments

02

Evaluation checks in CI

Bottom line

Consider it with an owner for examples and scorers; check cost meters, architecture and contract before relying on evaluation scores.

Key Features

Datasets and experiments
Code and model scorers
Evaluation playground
SDK and CI integration

Tags

AI evaluationExperimentsDeveloper tools

Compare Braintrust with alternatives

Frequently asked questions

Is Braintrust currently available?

No lifecycle review is recorded for this listing. Its inclusion in the directory does not confirm current availability; check the vendor before relying on it.

What is Braintrust?

Braintrust is a AI Platforms tool developed by Braintrust. AI evaluation platform for datasets, experiments and scorers, with hosted service and enterprise self-hosted data-plane options.

What costs and terms should I check for Braintrust?

Starter $0/month: $10 model credits, 1GB processed then $4/GB, 10,000 scores then $2.50/1,000, retention 14 days. Pro $249/month: $100 credits, 5GB then $3/GB, 50,000 scores then $1.50/1,000, retention 30 days then $0.50/GB/month up to 180 days. Enterprise quoted. Applicable paid terms specify USD, tax extra; regional totals and order billing remain unconfirmed. Model/Topics credits differ from score counts; own-provider keys have external fees. MSA defaults to annual advance invoicing unless the order differs; monthly list prices do not guarantee installment billing. Startup offers have separate eligibility.

Is Braintrust open source?

Apache-2.0 covers the SDK with notice obligations, not the proprietary hosted platform. The MSA preserves customer data ownership, permits service processing, and expressly excludes AI/ML training, development or improvement on Customer Data. It still permits nonidentifying Usage Data for improvement and retention under configuration/legal exceptions. Do not extend the MSA promise to every free account: the July 2026 general ToS permits service-operation content use without repeating that express no-training clause. Executed contracts take priority. External models, data and outputs retain separate rights and processing terms. The July 2026 ToS §3.1 limits Individual Users to personal non-commercial use and Business Customers to internal business use; a separately executed agreement governs conflicts.

What are the best alternatives to Braintrust?

Notable AI Platforms alternatives include Hugging Face, Replicate, Ollama. Browse all 138 tools in our AI Platforms category for a full overview.

Was this listing useful?

Visit official website →

Basic Info

Pricing
Freemium

Starter $0/month: $10 model credits, 1GB processed then $4/GB, 10,000 scores then $2.50/1,000, retention 14 days. Pro $249/month: $100 credits, 5GB then $3/GB, 50,000 scores then $1.50/1,000, retention 30 days then $0.50/GB/month up to 180 days. Enterprise quoted. Applicable paid terms specify USD, tax extra; regional totals and order billing remain unconfirmed. Model/Topics credits differ from score counts; own-provider keys have external fees. MSA defaults to annual advance invoicing unless the order differs; monthly list prices do not guarantee installment billing. Startup offers have separate eligibility.

Editorial score
★ 4.6 / 5 How we score
Platforms
WebAPI
Last verified
Not recorded