Note: This article pertains to Tines default AI - this does not apply to those who bring their own AI model into Tines. Custom AI models do not use any credits.
Tines AI features run based on credits available. All plans, including our community edition, receive an allocation of monthly credits by default.
Use of AI Features are subject to the terms set forth in the AI Addendum available at: https://www.tines.com/legal/
Included credits
All plans include a number of monthly AI credits. The number of credits depends on your plan tier. Credits do not roll over from the previous month. Paid editions of Tines can purchase AI credit add-ons (100K monthly credits) through your Customer Success Manager.
AI Features that use credits
There are two features in Tines that consume AI credits:
How credits are calculated
Credits are deducted based on the actual amount charged by AWS, not a fixed amount per execution. We are charged every time data is sent to the LLM provider and returned from the LLM provider.
AI Agent action (no tools) | Credits are deducted based on the volume of input and output data passed through the action - this is calculated each time the action runs |
AI Agent action (tools and/or chat mode) | Credits are deducted for each conversation entry (cycle of data input, tool use, etc.) |
Workbench | Credits are deducted for each message or tool use |
When you use the AI Agent action type without any tools (similar to our legacy AI action), you'll typically use fewer credits than in the past since we only subtract credits for what is consumed rather than a flat per-execution cost. Since agents can use much larger context windows and can run multiple cycles, complex agents can consume a significant number of credits.
To understand the costs we're incurring and get a relative idea of the cost per execution, review this AWS Bedrock documentation.
What if I run out of credits?
You trust us with your most important workflows, so we protect those workflows and provide warnings when you're nearing your credit limit. You can monitor your credit usage in Reporting and track usage by team in the AI settings modal.
Tines provides notifications when your team approaches credit thresholds. It is your responsibility to monitor usage and manage consumption to avoid disruption. Credit exhaustion may result in paused AI executions until credits are restored.
Available models
You can choose from multiple LLMs including Anthropic Claude models, Meta Llama 3 70B, and Amazon Nova models. We're expanding model support as AWS Bedrock adds new options. Models will differ depending on the AWS region that your Tines instance is hosted in.
Rate limits
For optimal performance, we rate limit by action runs per tenant per minute:
Tines AI provider: 40 runs per tenant per minute
Custom AI provider: 500 runs per tenant per minute
These rate limits are subject to change at Tines’ discretion to ensure system integrity.
Tip! To check where your tenant is hosted, append /info to your tenant URL.
The URL should look like this: https://acme-corp.tines.com/info