AI credits

Every AI operation in Crux debits AI credits from your allowance. Credits make AI usage predictable — you always know what you're spending and what's left.

How costs work

Each operation type has a published base cost — an Ask answer, the agent runs during project setup, a Generate on a detail field, or materialising an answer into notes, personas, journeys, tables, or to-dos. Larger operations cost more, and a turn with Web search on is metered as its own operation type. The debit happens only after a successful response; failed runs don't charge you.

Credits and the Agent model

The model you pick in the Agent composer is the one lever that changes what a turn costs, and it moves with the amount of thinking you asked for:

• Fast stretches your allowance furthest, and is the default for that reason.
• Sharp draws slightly more per turn.
• Capable draws noticeably more, because it is doing a deeper pass over more of your evidence.

Everything outside the composer runs on the managed default, so a Generate or a skill costs the same every time regardless of what the Agent is set to. A practical habit: stay on Fast while you are iterating, and switch up for the turn that actually needs it.

Allowances

• Sign-up grant — new accounts receive starter credits to explore.
• Trial — the free trial grants a Pro-sized credit pool for its duration.
• Monthly allowance — paid plans refresh a monthly allowance (larger on team tiers).
• Pooled team credits — organisation plans share one pool across all members.

Your current balance is visible in the sidebar's plan badge and in settings.

Running low

If you hit your allowance:

• Upgrade — higher tiers include larger monthly allowances.
• Request more — from the sidebar you can submit a usage increase request, which the Crux team reviews.

What doesn't cost credits

Editing, uploads, search, sharing, and the in-app Help agent don't consume your AI credit allowance.