Back to docs

Credits & billing

Plans, auto-recharge, credit expiration, cost control, and BYOK — including Claude Code and Codex.

Last updated May 27, 2026 · 4 min read

Zero charges by credits, not seats. A credit is a unit of work — one credit represents a fixed quantity of compute, regardless of who ran the task or which model handled it. Idle workspaces cost nothing.

Plans

Free — $0/month. 10,000 starter credits. No card required. Enough to set up a morning brief schedule, build a few skills, and run a handful of one-off tasks.

Pro — $20/month. 20,000 credits per month. Credits roll over one month. For individuals who use Zero daily.

Team — $200/month. 120,000 credits per month. Credits roll over one month. Shared workspace, central billing, admin controls. For small teams that share skills and schedules.

Top-up credits are available on every plan at $1 per 1,000 credits.

Auto-recharge

Pro and Team workspaces can turn on auto-recharge: pick a balance threshold and a top-up amount, and the workspace refills itself when the balance dips below the threshold. A common setting is “recharge $20 when balance drops below $5” — then forget about it.

Auto-recharge runs through Stripe on the card you used to subscribe. Turn it off any time on the Billing page.

Do credits expire?

  • Starter credits (Free plan) — expire 1 month after signup, and aren't replenished.
  • Monthly plan credits (Pro and Team) — expire at the end of the billing cycle in which they were granted. Each grant is usable only for that cycle.
  • Top-up and auto-recharge credits (any plan) — never expire.

The soonest-expiring credits are always consumed first. The Billing page shows your balance broken down by source, so you can see which bucket is closest to rolling off.

Cost control

The biggest knob is which model runs the work. Bigger models cost more — an Opus run can be 100× the cost of a Haiku turn. Set the default model per agent or per skill, so light triage runs on a cheap model and heavy research runs on a strong one. Workspace admins can also restrict which models are available to members.

Other knobs admins use:

  • Workspace cap. A monthly credit ceiling — Zero stops new chats when hit; admins can lift it mid-month.
  • Balance alerts. A Slack notification when the balance drops below a threshold you set.
  • Usage page. Burn rate, projected runway, top chats and top spenders for the last 7 days, broken down by member, skill, and schedule.

Bring your own key (BYOK)

Route inference through your own provider account instead of paying VM0 credits for tokens. Available on every plan, set up on the Model providers page:

  • Connect Claude Code. One click — Zero opens Anthropic's approval page, you sign in, and your existing Claude Pro or Max subscription is linked. No tokens to copy.
  • Connect ChatGPT (Codex). One click — Zero opens OpenAI's approval page, you sign in, and your ChatGPT Plus, Pro, Business, Edu, or Enterprise subscription is linked. No tokens to copy.
  • Anthropic / OpenAI / Google API keys. Paste a provider API key if you'd rather route through a direct API account than a subscription.
  • OpenRouter / Vercel AI Gateway / Bedrock / Vertex. For teams that already route inference through their own gateway.

When a chat uses BYOK inference, the token cost is billed to your provider account. Tool calls, connector usage, storage, and media generation still consume VM0 credits. The chat log shows the split per turn.

BYOK is workspace-level and can be assigned per model. Individual chats can override.

What's next