Plans

Harness has two plans: Open (free) and Resident (paid). Managed inference is always metered and available on either plan, so the plan you are on is about durability and depth, not about how much AI you can use.

Open

Free, forever.

• Run your own model locally, or bring a Venice or Bankr key.
• On-device memory that never leaves your machine.
• $5 of managed inference to start, top up any time.

Open is a complete product. Your memory and screen understanding are fully local, and nothing about Resident is required to use Harness every day.

Resident

Resident is the one paid plan. It sells durability and depth, not a bigger inference allowance.

• Everything in Open.
• Encrypted cloud backup, with a key only you hold.
• Device sync: pick up where you left off on any device.
• The full skills library.
• Image generation, with a private option.
• Early access to new features before they reach everyone, like the deep-agent brain for harder, multi-step questions.

Two ways to unlock Resident

• Pay monthly: $7.99 / month by card.
• Hold $HARNESS: hold the required amount of the $HARNESS token and Resident unlocks while you hold it. See Hold to unlock.

Both doors reach the same Resident. Whichever you use, you get the same features.

Managed inference and credit

Managed inference is pay-as-you-go for everyone, on either plan:

• Every account starts with a one-time $5 of managed inference free on signup.
• Top up in $10, $20, $50, or $100 packs, or pay with $HARNESS.
• Or bring your own key and skip managed inference entirely.

There is no recurring inference allowance bundled into Resident. The only free inference credit is the one-time $5 signup grant. Resident is about backup, sync, skills, image generation, and early access (like the deep-agent brain).