Free tier limits

The free tier exists so you can evaluate BrandBanta without a credit card. It's strictly gated to keep the shared OpenRouter key affordable for us to provide. None of the gates are arbitrary — each one maps to a real cost dimension.

Scan limits

Two different caps apply, and the stricter one governs. This trips people up, so read both rows:

The shared-key cap is a lifetime total, not a monthly allowance. A workspace with no BYOK key gets 5 scans ever. Once those are used, scans are blocked with FREE_TIER_EXHAUSTED until you add your own key — waiting for the next month will not give you more.

The 28-day tier quota is a separate counter that does reset. It only becomes the binding limit once you've added a BYOK key, because at that point your scans no longer draw on the shared-key lifetime cap.

The hourly cap is a rolling 1-hour window — if you ran a scan 65 minutes ago, you have full hourly headroom again (assuming lifetime headroom remains).

A third cap applies in shared workspaces, on any key. Members who can run scans but not manage API keys have their own per-member allowance, so one teammate can't spend the whole workspace's budget. It's independent of everything in this table and applies even after you add a BYOK key. See team collaboration. Owners, admins, and scheduled scans are exempt.

Resource limits

* Intended policy — retention windows are not yet enforced; today every tier retains full history.

You can create a 2nd topic on Pro by upgrading or adding BYOK (BYOK doesn't bypass the topic cap — only the scan cap).

What's gated on the shared-key free tier

Beyond the scan cap, the shared-key gates apply to every scan run on our key:

• Forced cheap models — we downgrade to budget-tier model variants on each platform (e.g. gpt-4o-mini instead of gpt-4o)
• Max output tokens forced to 600 — keeps each response bounded so a runaway LLM doesn't burn our budget
• LLM extractor skipped — sentiment, position, and context type are NOT extracted. You only get the deterministic regex-based mention count from pass 1.
• No AI narrative on /insights — the Executive Summary card shows "Add an OpenRouter key in Settings → API keys for AI narratives"
• No scheduled scans — manual scans only
• No alerts — the rules engine doesn't fire notifications for free-tier workspaces
• No webhook delivery — even if you wire up a webhook, deliveries are suppressed

These all unlock by either:

1. Adding a BYOK OpenRouter key — switches inference to your account; gates lift on the next scan
2. Upgrading to Pro — unlocks features regardless of key source (Pro still requires BYOK or shared-key budget for inference)

Error codes on free-tier exhaustion

When you exceed a limit you get one of these structured errors back from the API:

The error message includes the exact numbers ("used X / Y, remaining Z"). The UI translates these into friendly messages with an "Add BYOK key" or "Upgrade" CTA.

When BYOK doesn't bypass a gate

Some gates are tied to your tier, not your key:

• Topic count — Free tier max 1 topic regardless of BYOK
• Scheduled scans — Free + BYOK still can't schedule
• Webhook delivery — Free + BYOK still can't pipe notifications

The mental model: BYOK changes who pays for inference; tier changes what features you have access to. You can have either, both, or neither.

Cost expectations on free tier

Because we cap output tokens and use cheap models, a free-tier scan is cheap for us to serve. With search grounding enabled the realistic worst case is roughly $0.40 per free workspace, in total — which is precisely why the shared-key cap is a lifetime one rather than a renewing monthly allowance.

For comparison, a BYOK scan with full models costs ~$0.005 – $0.015 (on your OpenRouter account).

See also

• Pricing tiers — what each tier costs and unlocks
• BYOK with OpenRouter — how to add your key
• Scans & prompts — what each scan does