Model tokens measure shared AI input and output for scans, fixes and proofs. They are not a cash balance or a fixed number of scans. Free is granted once; paid allowances refill monthly, including on yearly billing. No metered overage is charged.
Extra seats are $10 a month, and each one adds capacity to the shared pool rather than carving it up per person.
What changes as the tier rises.
Every value below is the limit the product enforces, not a summary of it.
What each plan includes, compared
Free
Plus
Pro
Max
Allowance
Shared AI allowance
5,000,000 model tokens once
12,000,000 model tokens per month
28,000,000 model tokens per month
65,000,000 model tokens per month
Refills
Never
Monthly
Monthly
Monthly
Scale
Repositories
2
10
Unlimited
Unlimited
Seats included
2
5
10
25
Extra seats
Not available
$10 a month
$10 a month
$10 a month
Scanning
Scheduled scans
Manual only
Daily
Every 6 hours
Hourly
Scan every push
Not included
Yes
Yes
Yes
PR source scans (changed files only)
Not included
PR source scans (partial coverage)
PR source scans (partial coverage)
PR source scans (partial coverage)
Max depth
Not included
Not included
No run cap, shared token allowance
No run cap, shared token allowance
Container images
Not included
Not included
Yes
Yes
Platform
Immunity Watch
Not included
Not included
Not included
Yes
Hyper immunity tests
Not included
Not included
Not included
Yes
SSO and audit log export
Not included
Not included
Not included
Yes
What the plan actually means.
The answers that decide whether a plan fits, stated before you pay for it.
One organization. Cefense bills per organization, the way Vercel bills per team and GitHub bills per organization. The organization is the tenant, the payer, and the meter. An account can hold a paid organization and a free one at the same time, and each is metered on its own.
No. Free is a one-time allowance, spent once rather than renewed each month. It is not a trial either: the organization, its repositories, and everything already scanned stay. What runs out is the ability to start another scan.
New scans pause, with no metered overage. Existing findings remain accessible. Free does not refill. Paid allowances refill monthly, including on yearly subscriptions. A plan change can add allowance; review its price before confirming.
Model tokens measure the AI input and output used by scans, fixes and proofs. They are usage units, not a dollar balance. There is no fixed tokens-per-scan conversion: repository size, selected coverage checks and scan depth change consumption. Billing shows the live shared meter and the applicable refill date.
The organization does. Every model call a scan, a fix, or a proof makes is drawn from one shared pool. Seats add capacity to that pool rather than dividing it, so a quiet month on one repository pays for a loud one somewhere else.