Zia, the AI assistant in Zoho CRM, can connect to external AI models or to models hosted by Zoho, so their output flows into your records, workflows, and conversations. Every time one of these models runs it uses tokens. That is true whether you use the Zoho Hosted LLM, the Zoho Key Service (ZKS), or your own API keys. On the Zoho Key Service, models also use up credits.
The AI Models page in Zoho CRM now has two new tabs, Usage and Limits. They let administrators see where tokens and credits are going, and cap consumption before it gets out of hand.
Availability: Zoho's announcement does not state which editions, data centres, or rollout dates apply, so check your own CRM account's AI Models page before promising this to a client.
The problem it solves
- CRM used to show only a shared credit balance. For anything more detailed, you had to check each vendor's own dashboard.
- Nothing in CRM told you which vendor or model was driving your spend.
- There was no way to stop a model once usage crossed a threshold, so if one team started consuming heavily, the cost could climb quietly until someone noticed.
The Usage tab: where your tokens and credits go
Start by choosing a Model Connection Type. The view changes depending on your choice.
| Connection type |
What the view shows |
| All | Token consumption across every connection type. |
| Internal | Token consumption for Zoho-hosted models. |
| BYOK (bring your own key) | Token consumption for models you connect with your own API keys. |
| ZKS (Zoho Key Service) | You can switch the whole view between tokens and credits, because ZKS is the credit-based option. |
The headline numbers
Four figures sit at the top of the tab: input tokens, output tokens, the top vendor, and the top model. Together they show at a glance where most of your spend goes.
The three charts
- Token Usage Trend. Plots input and output tokens over time for each connection type, so spikes are easy to spot.
- Model Usage Comparison. Puts consumption side by side, by vendor or by model.
- Model Usage Trend. Tracks consumption for each model over a period you choose: today, the last 2, 7, 14, or 30 days, or a custom range.
The Limits tab: a ceiling on consumption
The Limits tab lets you cap token consumption before it becomes a problem.
- One vendor and one profile per rule. Each pairing of vendor and profile must be unique, so it is always clear who a limit applies to.
- As many rules as you need. You can switch any rule on or off from the list, and edit or delete it later.
- On a schedule. Limits can run on a daily, weekly, or monthly basis.
- Applied per user. Once a user reaches the limit, Zia stops running models from that vendor for them and asks them to contact their administrator for more. Because the limit applies to each user, one heavy user cannot quietly drain a shared balance.
Why it matters
As more teams in your CRM use AI, the cost of running these models gets harder to see as it happens. The Usage tab puts the numbers on the same page where you manage your models, so you can decide which models to keep, which to adjust, and which to retire. The Limits tab lets you set boundaries that fit your budget and your team's needs, per vendor and per profile.
Need help? As a Zoho AI consultancy and Zoho Partner serving the UK and Ireland, we can help you choose which models to run in Zoho CRM, set sensible token limits by profile, and keep AI costs predictable.
Book a discovery call with 1 Cloud Consultants.