When you run out
What stops when the allowance is used, what keeps working, and what to do about it.
WARNING
There are no paid plans today. This describes functionality that is built but not launched.
Running out of allowance is a normal state, not a failure. Here is exactly what changes and what does not.
What stops
Only one thing: model requests routed through the NALA gateway.
What keeps working
Everything else, which is most of NALA:
- Your own API keys. Requests go straight from your machine to the provider. They do not pass through our servers, so nothing we do can stop them.
- Local models. No account, no network, no gateway.
- Third-party agent harnesses. NALA launches the vendor's own CLI, which authenticates itself.
- The whole application — terminals, orchestration, worktrees, A2A, everything that is not a NALA-provided model call.
This is not a courtesy. It is structural: the free product does not depend on our servers, so it cannot be taken away by a billing state or by our infrastructure failing.
Your options
| Wait | The allowance resets every 7 days. nala usage shows exactly when. |
| Use purchased credits | If you have any, they are used automatically once the allowance is gone. |
| Buy credits | These do not expire. |
| Use your own key | Works immediately, costs you nothing extra through us. |
| Use a local model | Free, offline. |
If the gateway is down
Same picture, for a different reason. Your own keys and local models are unaffected, because those requests never involved us.
If a request fails because our gateway is unavailable, you are not charged and any credits reserved for it are released.
What we will not do
- Charge you automatically without a ceiling you set. Automatic top-up is off unless you enable it, and enabling it requires an upper bound.
- Stop the application working. There is no state in which running out of credits disables NALA.
- Retroactively bill you for a run we under-charged. That is our accounting error to absorb.