AI usage, explained
Three pools. No surprises.
Sentiment + Subjects, Moderation and Chat have separate monthly token allowances. Using one never reduces the included allowance of another.
Sentiment + Subjects tokens
Understand what people discuss and how they feel. Sentiment and Subjects share one analysis and one charge.
Moderation tokens
Review borderline messages and apply your written server rules. Basic screening consumes no tokens.
Chat tokens
Ask Zero, explore your data and generate insight cards. Longer answers and multi-step work can use more tokens.
Your monthly allowances
| Included each month | Standard | Pro |
|---|---|---|
| Subscription | $29/month | $79/month |
| Billed annually | $290/year | $790/year |
| Sentiment + Subjects tokens | 30,000 | 150,000 |
| Moderation tokens | 20,000 | 30,000 |
| Chat tokens | 15,000 | 75,000 |
| Illustrative sentiment capacity | About 2,000 messages | About 10,000 messages |
| Optional additional usage | $1 per 1,000 tokens in any pool. One shared spending cap. | |
Message estimates assume 15 Sentiment + Subjects tokens per message. Actual usage varies with context, batch size and AI work. These are examples, not guaranteed message allowances. Sentiment and Subjects are analyzed together and charged once.
Free includes 500 Chat tokens each month. Sentiment + Subjects and AI moderation start on Standard.
Tokens measure work, not messages
Metryx tokens are billing units, not the text tokens used by AI models. All three pools use the same cost-based rate. A simple request can use fewer tokens than a long conversation or a detailed analysis.
See your balances and recent daily usage in Settings → AI usage.
You control additional usage
Additional usage is off until you enable it. Then it costs $1 per 1,000 tokens, with one monthly spending cap shared across all three pools. If a pool runs out, that AI work pauses. Your other included pools and basic moderation screening keep working.
Fresh allowances every month
All pools reset on the first of each month at midnight UTC, including on annual plans. Unused included tokens do not roll over.
Failed work is not a customer charge
We release reserved tokens when an operation fails. A temporary reservation may appear while a request is running. Turning off additional usage stops new paid reservations; usage already authorized may still settle within your existing cap.