Credits and pricing
AI FAQ uses your server's AI credits every time the bot generates an answer or you save a new FAQ entry. This page covers what each thing costs, how your monthly grant works, and what to do when credits run out.
Monthly free grant
Each paid plan tier comes with a free monthly grant of AI credits, topped up on the 25th of each month. Unused credits roll over into the next month - you never lose what you didn't spend.
| Plan tier | Monthly credits |
|---|---|
| STARTER | 25 |
| ACTIVE_GUILD | 100 |
| PRO | 250 |
| UNLIMITED | 300 |
| PROFESSIONAL | 600 |
If you regularly run out before the monthly refresh, you can top up from the Pricing page in your dashboard.
What each answer costs
Every AI answer charges a small amount of credits. The exact cost depends on what the AI actually had to do:
- A rejected message costs 1 credit. This happens when the AI refused to answer (e.g. abusive content) or the question was filtered out before it ran.
- Small talk or off-topic chatter costs 3 credits. This happens when the AI handled a quick "thanks" or "hi" without needing to search your FAQ.
- A real FAQ answer costs 15 credits. This is the most common case: the AI searched your FAQ entries, found the right one(s), and wrote an answer based on them.
Provider multiplier
The numbers above are for the OpenAI provider, which is the cheapest. If your server is set to a different provider, the per-answer costs are multiplied:
- OpenAI - base rate (1×)
- Google Gemini - 2× the base rate
- Anthropic - 3× the base rate
So a real FAQ answer on the Anthropic provider works out to 45 credits. Your current provider is shown in Settings → Status on the AI FAQ page, along with how many answers your current balance will buy you at the current provider's rate.
What FAQ entries cost to save
When you save a new or edited FAQ entry, the bot splits it into searchable chunks so the AI can find the right one later. Each chunk costs 1 credit to index.
- For typical English prose, a chunk works out to roughly 2,000 characters - so a 4,000-character entry costs about 2 credits to save, and a 20,000-character entry around 10 credits. Code-heavy content tends to chunk smaller, so it can be a little higher per character.
- The entry editor shows the exact chunk count and credit cost live as you type, so you can see what it'll cost before you click save.
- Editing just the title, or archiving/unarchiving an entry, is free. Only changes to the answer text trigger the recharge.
- Re-saving without changing the answer is also free.
- Archived entries don't cost anything to keep around.
See Writing FAQ entries for the same thing in context of the editor.
Realistic monthly spend
Three rough sketches of what a month looks like at different volumes. Real numbers depend on the mix of question types (rejected vs. small talk vs. real search) and your AI provider.
Quiet help channel, mostly small talk
A small community where the AI sees around 30 questions a month, most of which are short chatter or "thanks" rather than real FAQ lookups.
- ~20 small-talk interactions × 3 credits = 60 credits
- ~10 real FAQ answers × 15 credits = 150 credits
- Total: roughly 200 credits / month
Easily covered by the PRO (250) or UNLIMITED (300) monthly grant. ACTIVE_GUILD (100) would need a small top-up most months.
Active support server, OpenAI
A busy support server where the AI handles ~150 questions a month, mostly real FAQ lookups via the channel auto-answer.
- ~30 small-talk × 3 credits = 90 credits
- ~120 real FAQ answers × 15 credits = 1,800 credits
- Total: roughly 1,900 credits / month
Well above any plan's monthly grant - you'll be topping up regularly. If you're consistently here, consider talking to us about enterprise volume pricing.
High-traffic server on Anthropic
A larger server using the Anthropic provider (3× multiplier) with ~500 questions a month, mostly real FAQ lookups.
- ~100 small-talk × (3 × 3) = 900 credits
- ~400 real FAQ answers × (15 × 3) = 18,000 credits
- Total: roughly 19,000 credits / month
Well beyond plan grants. Either switch to OpenAI to drop the multiplier (saving roughly two-thirds), or move to an enterprise plan with volume-priced credits - see Enterprise & Docs Sync.
Topping up credits
Credit top-ups are available on the Pricing page in your SCNX dashboard. Top-up credits stack on top of your monthly grant and roll over until you use them.
Top-up rates are flat per credit. If your server regularly spends well above its monthly grant, volume-rate pricing and fixed monthly budgets are available on enterprise plans - see Enterprise & Docs Sync or contact sales.
When credits run out
When your server's balance hits zero, the AI can't answer anything until you top up. You control what members experience during that window under Settings → Fallback behavior → When credits run out:
- Tell members the server is out of AI credits (recommended, default) - the bot sends a short note explaining the AI is temporarily unavailable. Keeps the member in the loop and gives staff a visible signal to top up.
- Offer to open a ticket instead - the bot shows a button so the member can still reach staff. They aren't told why.
- Stay silent (do nothing) - the bot says nothing at all. Members might think the bot is broken. Only use this if you actively monitor your balance.
Cutting credit spend
If you're going through credits faster than you'd like, the biggest knobs are on the channel auto-answer side - the AI runs much more often there than in the gatekeeper:
- Raise the Cooldown so one member can't trigger answer after answer in quick succession. See channel auto-answer filters.
- Raise the Minimum question length so short chatter is skipped.
- Add more prefixes to ignore to skip command-style messages from other bots.
- Disable channel auto-answer on noisy channels - leaving the pre-ticket gatekeeper running on its own is usually much cheaper, because it only fires when someone is actually opening a ticket.
- Archive outdated FAQ entries so they don't take up space in the AI's search.
- Switch to a cheaper provider if your current provider's quality difference doesn't justify the multiplier.
The Settings → Status panel on the AI FAQ page shows your current balance, roughly how many answers it'll buy you, and your current provider. The Insights page shows how many answers have actually been generated and where they went.