Skip to main content
Three limits apply: a free allowance of 3 merged deliveries before you subscribe, a monthly token allowance that depends on your paid plan, and a limit on how many repositories can be connected at once. None of them ever interrupts a delivery that is already running.

​
At a glance

Prices and the yearly option are on the pricing page; billing covers subscribing.

​
The free allowance

Your first 3 merged deliveries are free. No card, no subscription. A slot is taken when you label an Issue ai-ready, and it is refunded if that delivery does not end in a merge: an Issue that ends ai-blocked, that Orbi sends back as needing more detail (ai-needs-detail) before starting, or that you close without merging, gives its slot back. That is why the allowance is described as three merged deliveries rather than three attempts. The count does not reset monthly. Counting is per Issue and idempotent: re-labelling the same Issue does not consume another slot. Re-delivering the same Issue after a refund takes a slot again. The status page shows the remaining free allowance in its usage summary. Setup has five steps and finishes when you hand over the first request; it does not end with a subscription step. A verified official example copy is separate: its deliveries do not consume your three free deliveries, and one connected copy does not use a repository slot. This exemption applies to the official template copy, not to an arbitrary repository name.

​
When the allowance runs out

The next Issue you label ai-ready has that label removed and gets a comment saying the free deliveries are used up, with a link to subscribe. The status page says Your 3 free deliveries are used up: subscribe to keep delivering → with a Subscribe button. To resume, subscribe, then re-apply the ai-ready label to the Issue. The label is not restored automatically. Deliveries already in flight when the allowance ran out finish normally.

​
The monthly token allowance

A paid subscription includes model usage: 400,000,000 tokens a month on Solo, 1,200,000,000 on Pro, counted per calendar month in UTC. The status page shows the reset as the 1st at 8:00 Beijing time, which is the same moment as 00:00 UTC. There is no per-token overage billing. Free accounts see no monthly allowance; their limit is the free deliveries. Usage counts input tokens, output tokens and cache-read tokens, all at the same weight. Re-delivering the same Issue accumulates: each delivery is a separate run. How many deliveries that buys depends on how large the tasks and the codebase are, and on prompt caching. The pricing page gives current measured ranges. Model usage includes deliveries, drafting requests and generating suggestions. For each, input, output and cache-read tokens count toward usage. Platform usage and usage on your own key are recorded separately.

​
Watching your usage

Open Usage and settings on the status page. On a paid plan the Monthly AI usage line shows tokens used against your allowance, the Used percentage, the Cached share and when it resets, with a bar that turns amber at 80% and red at 100%. On the free plan the line shows only the tokens used, with no cap, percentage or bar. Cached is the share of this month’s tokens that were cache reads. Cache-read tokens count against the allowance just like input and output tokens. If some tasks did not report their usage, the line says how many were left out of the total.

​
When the monthly quota runs out

Orbi pauses claiming new Issues. Your ai-ready labels stay where they are, and deliveries already running finish normally. The main status card says Monthly AI usage is used up: Orbi has paused new tasks; in-progress ones will finish normally. and offers the ways out:
  • Use your own API key. Saving a key in the model configuration resumes delivery immediately, does not count against this month, and is billed by your AI model provider. See model configuration.
  • Upgrade to Pro (shown on Solo), for 1,200,000,000 tokens a month.
  • Or wait. On the 1st of the next month at 8:00 Beijing time (00:00 UTC) the allowance resets and the waiting Issues start automatically. You do not need to re-apply any label.

​
The repository limit

Free and Solo connect one repository at a time; Pro connects up to five. The rule is the same on every plan: at the limit, connecting another repository is refused with Repository limit reached, so disconnect one first with Disconnect a repository, then connect the new one. Disconnecting does not change the repository on GitHub. Free and Solo can also upgrade to Pro. See connect a repository.

​
What is never interrupted

No limit ever stops work in progress. A delivery that is ai-in-progress, ai-pr-opened or ai-fix-needed runs to completion regardless. Platform upgrades and deployments have the same property: they never restart a delivery that is already executing.

​
Summary