CiteTrack tracks citations by asking each AI engine your questions through your own API key. You pay each provider directly, at cost, with nothing marked up. Because the keys are yours, the spend is small and entirely in your control. Here is the honest math.
What one check actually costs
Each check is a single API call: one tracked prompt sent to one engine, using that provider’s supported web-search or grounding tool when available so the answer can include current cited sources. The exact price changes by provider, model, tokens and search/tool usage, so the plugin now keeps a local cost ledger instead of relying on one fixed public estimate.
| Provider | What CiteTrack records | How to read it |
|---|---|---|
| OpenAI / ChatGPT | Model, token usage, tool/search usage and estimated dollars when available. | Good for ChatGPT visibility, citation and answer-monitoring checks. |
| Google Gemini | Model, grounding/search usage and estimated dollars when available. | Useful for low-cost broad monitoring and comparison against other engines. |
| Perplexity | Model, search/citation usage and provider metadata. | Strong fit for source-heavy answer checks. |
| Anthropic / Claude | Model, tokens, tool usage and estimated dollars when available. | Useful when buyers in your market use Claude for research. |
Estimate your own bill
Your monthly cost is simply prompts × engines × how often you check. Weekly checks run four times a month; daily checks, thirty. Try your numbers:
List prices, June 2026. Cost is driven mostly by each provider’s per-search fee, not tokens, so cutting engines or frequency lowers it far more than switching models. Your real bill varies with answer length; set exact rates in Settings → Costs.
What that looks like in practice
| Who | Setup | Calls / month | Cost |
|---|---|---|---|
| Blogger / small business | 10 prompts, 2 engines, weekly | 80 | ~$2 |
| Active SMB | 25 prompts, 3 engines, weekly | 300 | ~$6–9 |
| Agency | 50 prompts, 4 engines, daily | 6,000 | ~$150–180 |
Most site owners land in single digits a month. The agency figure is high only because it checks fifty questions against every engine every day, a cadence you choose, not a floor.
How to keep it low
- Check weekly, not daily. AI answers move slowly; weekly catches the trend at a seventh of the cost.
- Track the prompts that matter. Ten high-intent buyer questions beat a hundred vanity ones.
- Start with one or two engines. Add the rest once you see the value.
- Use the cost ledger. Review the actual provider/model mix instead of guessing from averages.
- Set a budget at the provider. Every API console lets you cap monthly spend, so there are never surprises.