Planned pricing

Start free. Scale when it helps.

Every plan includes the repo-grounded workflow. Choose by how often you delegate work and how much control you need.

Access by request

Public billing is not open. Requesting access does not start a subscription or collect payment details.

Request access

Compare plans

Choose by how you work.

Prices shown are monthly. Planned annual billing is 20% lower and billed once per year.

ExploringStart with Free
Coding most daysChoose Professional
Running agents heavilyCompare Engineer and Power

Free

Explore the core workflow

Monthly price

$0/mo

Light monthly usage

  • Terminal agent
  • Repository context
  • Bring your own API key
Request Free

Hobby

Personal and side projects

Monthly price

$10/mo

$8 with annual billing

Regular personal use

  • Tool use and planning
  • Code review
  • Personal repositories
Request Hobby

Engineer

Frequent multi-agent work

Monthly price

$50/mo

$40 with annual billing

About 7x Hobby capacity

  • Custom MCP tools
  • Priority models
  • Advanced routing
Request Engineer

Power

Maximum individual capacity

Monthly price

$100 / $200/mo

$80 / $160 with annual billing

About 15x or 35x Hobby capacity

  • Highest-priority routing
  • Long, parallel sessions
  • Heavy automation support
Request Power

Prices, usage limits, model access, and billing terms remain subject to change until public billing opens.

Before you choose

Common questions.

Can I buy a plan today?

Not yet. Public billing is not open. The prices and limits on this page describe the planned model, and final terms will be shown before purchase.

What do the usage multipliers mean?

They compare included agent usage with the planned Hobby baseline. They are not token allowances. Exact limits will be published before paid access opens.

What if I need more usage?

The planned options are moving to a higher-capacity plan or adding prepaid, token-based usage without changing plans.

Is repository data used to train models?

No. BuildStax does not use repository data to train models. Task context may be sent to selected providers for inference under their applicable data terms.