Tier limits
Every paid resource is metered against the organization that owns the API key. Limits are enforced per organization, not per user.
At a glance
Section titled “At a glance”| Plan | Concurrent sessions | Monthly browser-minutes | AI tokens/month |
|---|---|---|---|
| Free | 1 | 100 | 50,000 |
| Starter | 5 | 1,000 | 500,000 |
| Team | 25 | 10,000 | 5,000,000 |
| Business | custom | custom | custom |
What counts
Section titled “What counts”- Browser-minutes — wall-clock time a session is alive. Idle time inside a session counts; pooled-but-detached time does not.
- AI tokens — used by Slice E’s AI-step exploration when it ships. Slice C-MVP does not consume AI tokens.
- Concurrent sessions — peak simultaneous live sessions across the org.
What doesn’t count
Section titled “What doesn’t count”- Test-mode sessions ARE metered against your plan (no free pass), but capped at 5-minute max-duration so they can’t spike usage.
- Tool calls themselves are free; only the underlying browser time is metered.
- Health checks (
kifas.health) are free.
What happens at the cap
Section titled “What happens at the cap”- Concurrent-session cap reached —
kifas.session.createreturnsforbidden(403). Existing sessions continue. - Monthly browser-minute cap reached —
kifas.session.createreturnsforbidden(403) until the next billing period or a plan upgrade. Existing sessions are NOT killed. - Rate limits (per-key, per-minute) — short bursts return
rate_limited(429) withRetry-After. Backoff in your client.
Pricing
Section titled “Pricing”Full pricing lives at kifas.io/pricing. Billing runs through Paddle (Merchant of Record), so VAT/sales tax is handled in every supported jurisdiction.