How Braintrust prices
Captured September 19, 2026 at 1440px wide.
Braintrust plans and prices
| Plan | Price | Billing |
|---|---|---|
| StarterFor everyone | $0 | / month |
| ProFor AI native teams | $249 | / month |
| EnterpriseFor teams at scale | Custom |
As printed on the page captured September 19, 2026.
How does Braintrust pricing work?
The value metric is consumption. Starter is $0 per month with $10 credits, 1 GB processed data, 10k scores and 14-day retention. Pro is $249 per month with $100 credits, 5 GB, 50k scores and 30-day retention. Enterprise is 'Custom pricing'. Every allowance prints its marginal rate in the same line: '1 GB processed data + $4/GB', '50k scores + $1.50/1k', and '30-day retention + $0.50/GB/mo'. Users, Projects, Playgrounds, Experiments and Datasets read Unlimited on all three tiers. For an eval platform the cost driver is how much traffic a team traces and scores, so the meters sit on data and scores and the whole team can work in one workspace.
The motion is self-serve through Pro and sales-led at Enterprise, which adds 'Custom data retention and export, RBAC, and premium support with on-prem or hosted deployment for high volume or privacy-sensitive data.' and routes to 'Contact us'. The page opens on 'Predictable pricing. Designed to scale.', with a left-hand column of row labels, Model credits, Processed data, Scores, Data retention and Features, that turns the three cards into a matrix. A logo band follows, then a repeated plan header with live CTAs over a comparison table grouped into Core platform, Customization, Data retention and export, Security and compliance, and Support, each with a one-line purpose. A usage calculator, six FAQ questions and a closing 'Trace everything' band with Sign up complete the page.
Under the hood the bill has two layers. The platform layer is a fee plus included allowances on processed data, scores and retention, each with a printed overage rate. The model layer is credits, '$10 credits / month included then token rates' on Starter and $100 on Pro, drawn down at per-model token rates on a separate rate card that 'View detailed pricing, including model and token rates' links to. That keeps pass-through model spend apart from the value Braintrust adds as a platform. The usage calculator reflects the same split: it converts tokens per month and a scoring percentage into traces, GB of storage and scores, and prices the platform meters on each tier. Pricing what the platform does in its own units, and model usage at model rates, is a structure worth borrowing for any AI tool that sits on top of third-party models.
What does Braintrust's pricing page do well?
- Every included allowance prints its overage rate beside it on the card: +$4/GB, +$3/GB, +$2.50/1k scores, +$1.50/1k scores and +$0.50/GB/mo for retention
- Users, projects, playgrounds, experiments and datasets are Unlimited on all three tiers, so a whole team can instrument its AI stack at no extra cost
- A left-hand label column turns three cards into a matrix, so each card carries numbers and the reader scans one dimension across all tiers
- Pro carries a fixed $249 / month platform fee, a clear number to budget against
- 'No credit card required' sits under Sign up and '6–12 months free for qualifying startups' sits under Get Pro, each answer placed at the button it serves
What does Braintrust's pricing page reveal about its pricing strategy?
- Two billing layers: a platform fee with metered data, scores and retention, and model credits drawn down at token rates
- Keeping token rates on their own rate card separates pass-through model spend from the platform value Braintrust prices
- Unlimited seats on every tier fit a product whose cost scales with traced traffic, and they encourage the whole team to adopt it
- The startup offer attaches to Pro, so early-stage teams land on the plan Braintrust wants them to grow on
- Enterprise adds on-prem or hosted deployment and custom retention, the features that set the cost of serving a large account and that the sales team prices deal by deal
Learn with Orb
- ProductExperience Kit A usage calculator turns tokens and scoring into an estimate, and every allowance prints its overage rate; Orb's Experience Kit builds calculators and dashboards so customers can plan spend.
- ProductPrice modeling A $249 platform fee, metered data, scores and retention, and model credits at token rates add up to a hybrid plan; Orb's price modeling covers logic like this without re-engineering billing.
Questions about Braintrust pricing
How much does Braintrust cost?
Braintrust's pricing starts at $249/mo. The page shows 3 plans, including a free plan, with the top plan priced through sales.
What pricing model does Braintrust use?
Braintrust prices by usage, combined with flat-rate tiers and credits. Buyers can sign up on their own or go through the sales team. Its plans are aimed mainly at developers.
Does Braintrust have a free plan?
Yes, Braintrust has a free plan. Paid plans start at $249/mo.
Does Braintrust offer an annual discount?
Braintrust's pricing page shows a single billing period. The listed prices start at $249/mo. Terms for the plans priced through sales are set with the sales team.
How is Braintrust's pricing page laid out?
The page shows 3 plans and includes a comparison table, a calculator and an FAQ. The breakdown above marks 14 points on it: 8 things it does well and 6 insights into the pricing.
What happens when I go over my Braintrust usage allowance?
Each Braintrust plan includes a monthly allowance and lists a printed overage rate for usage beyond it. Starter includes 1 GB of processed data "then +$4/GB" and 10K scores "then $2.50 per 1,000". Pro includes 5 GB "then +$3/GB", 50K scores "then $1.50 per 1,000", and 30-day retention "then $0.50 per GB per month". Enterprise sets these allowances as Custom.
How do Braintrust model credits work?
Starter comes with "$10 credits / month included then token rates", and Pro comes with "$100 credits / month included then token rates". Enterprise is listed as "Custom then token rates". The page links to "View detailed pricing, including model and token rates" for the per-model rate card.
Does Braintrust offer a discount for startups?
Yes. The Pro plan, normally $249 / month, carries the note "6–12 months free for qualifying startups" under its Get Pro button. The page footer also lists a "Startup program for fast-growing AI teams".
From the page as captured September 19, 2026. Current prices are on braintrust.dev
Alchemy
$0.52/1M CUAlchemy prices blockchain infrastructure on one published rate, $0.52 per 1 million compute units, and sells peak throughput as its own meter: $160 for each 5,000 CU/s block above the 300 RPS that Pay as You Go includes. A working estimator itemizes compute, included capacity and overage into a monthly total, so a developer can model the bill before signing up.
Augment Code
$20/mo flatAugment Code sells a flat team subscription with usage inside it: Standard is '$20/month flat, no per-seat charge' and Business '$100/month flat', each covering up to 50 seats and loading the same dollar amount of usage every month. Usage is measured in dollars, and the FAQ publishes the formula: LLM inference at 'the provider's public API list price', 'a flat 40% fee on LLM usage (no fee on compute)', and Cosmos compute time. The plan is a prepaid wallet shared by the whole team.
LaunchDarkly
$10/Service Connection/moLaunchDarkly prices the software it serves rather than the people who use it: Developer is '$0 / mo, forever' with 'Unlimited seats', Foundation is pay as you go with 'No platform or seat fees', and Enterprise is custom with a trial. Foundation meters three things, '$10 per Service Connection / mo', '$8.33 per 1k client-side MAU / mo' on yearly billing and '$5 per additional 1k runs' of AI, and the comparison table spells out what happens at each plan's limit.
Mux
$20/moMux prices video by the minute and teaches the unit on the page: a 'What the minute?' explainer, a scenario switcher for what $100 of credit buys, a full rate sheet and an inline estimator with SHOW ME THE MATH. Every plan includes 100K free delivery minutes and 100K Mux Robots units a month, resolution and quality work as multipliers on one base rate, and a dedicated 'AI workflows (Mux Robots)' list prices thirteen AI jobs by minute, job and, for generated text, artifact.
Neon
$0.106/CU-hourNeon prices serverless Postgres on raw resources, $0.106 per CU-hour and $0.35 per GB-month on Launch, and turns those rates into a budget with a Typical spend figure on each paid card, set by a workload dropdown. Compute scales to zero when idle, so the meter follows real work, and the Free plan states every allowance per project. A rewritten 'Included in every Neon project' section and new FAQ answers on Free plan limits, Claimable Neon and idle compute frame the backend around the database as part of every plan.
Railway
$5/mo minimum usageRailway bills compute by the second and wraps that meter in plans defined by a minimum usage floor that comes back in full as credits: '$5 minimum usage' on Hobby and '$20 minimum usage' on Pro. Further down, an 'Unlock more as you grow' ladder turns enterprise features into published monthly commitments, from $1,000 for HIPAA BAAs to $10,000 for Dedicated VMs.