Last updated: 19 September 2026 · Version 2.0
Every paid plan includes a monthly AI allowance, shared across all of Koda's AI features. The short version: use them as much as you need to run your business. The allowances below are sized so that normal agency work never runs out, and they exist because every AI request costs us money with our provider.
Koda does not meter each feature separately. Your plan has one pool per calendar month, and everything that calls the AI draws from it: captions, hashtags, content ideas, brand intelligence scans and refreshes, website scans and spell-check.
The pool is measured in tokens — the unit our AI provider bills in. A token is roughly three-quarters of a word, counted across everything the AI reads and writes. That includes the client's brand profile, which is re-read on every request so the output sounds like that brand. A client with a rich brand profile and uploaded brand documents therefore uses more per caption than a sparse one.
You can see exactly where you are at any time: Settings → Organisation shows your allowance, what you have used this month, and the percentage remaining.
| Plan | Monthly allowance | Most in one day | Guide: captions / month |
|---|---|---|---|
| Free | 100,000 tokens | 35,000 | around 65 |
| Personal | 300,000 tokens | 105,000 | around 200 |
| Freelancer | 500,000 tokens | 175,000 | around 330 |
| Studio | 3,000,000 tokens | 1,050,000 | around 2,000 |
| Agency | 8,000,000 tokens | 2,800,000 | around 5,300 |
The caption column is a guide, not a quota. We don't count captions — we count tokens. It assumes a typical client and around 1,500 tokens per caption; a client with large uploaded brand documents can cost several times that, and other AI features draw from the same pool.
The daily figure is a safety cap, set at 35% of your monthly allowance. It exists so that a runaway loop or a compromised account can't burn a whole month in an afternoon. Catching up on a backlog will not hit it; scripted abuse will.
At 80% of your monthly allowance we email the organisation admin, once per month, so there are no surprises.
At 100% new AI requests pause until the next calendar month resets the counter, or until you upgrade — whichever comes first. Everything else in Koda keeps working normally: scheduling, publishing, approvals, the inbox, analytics and reports are unaffected, and monthly client reports are genuinely unlimited because they use no AI.
We never charge overage fees. The only consequence of reaching the allowance is the pause.
Staying inside your allowance is not abuse, whatever the volume. These patterns are:
If we detect abuse we'll contact the organisation admin first. Continued abuse may result in account suspension under our Terms of Service.
This is what Koda is for. Use it.
If we change the thresholds, we'll email all active customers at least 30 days before the change takes effect. We commit not to lower a threshold without that notice.
If you're hitting a limit and don't think you should be, or if you have an unusual high-volume use case, email [email protected]. We'd much rather have a conversation than enforce a cap.