How credits work
Every action your AI threads take costs credits: thinking, processing data, sending a message, reading an email, running code, browsing the web, and more. This includes both actions you initiate manually and actions taken by automations.
You can purchase credits in any $5 increment, with a $25/month minimum. See our pricing page for options. We bill for credits at a price very close to our underlying costs to give you the best value possible.
The main component of your paid plan is a monthly credit allowance that renews each billing cycle. In addition, each day you use Tasklet, you receive a daily bonus to help cover personal daily usage (300 credits/day for free users, 600 credits/day for paid users). Daily bonus credits reset at Midnight UTC.
There are two ways to purchase credits:
- Monthly subscriptions include a recurring credit allowance that renews each billing period.
- One-time top-ups add extra credits whenever you need them.
Monthly subscription credits reset each billing cycle and do not roll over. Top-up credits remain available for 12 months from the date of purchase.
What affects credit usage
The credit costs of different AI queries can vary widely. The main factors affecting credit usage are:
- Intelligence level: Higher intelligence levels cost more per action.
- Task complexity: More steps and more data to process require more credits.
- Browser use: Browsing websites can use more credits. Prefer built-in connections for services when available.
- Automation frequency: Every automation run costs credits at the same rate as a manual action.
Making the most of your credits
The biggest lever is reducing how often AI gets invoked:
- Reduce automation frequency. One of the most common ways credits are wasted is running automations too often.
- Add automation filters. Many automations let you filter events before invoking AI (for example:
only process emails not labeled “Promotions”, or only run a webhook automation when
body.status === 'success').
You should also consider:
- Lowering intelligence levels for threads that do not need higher reasoning effort
- Starting fresh threads for unrelated tasks instead of continuing long threads
Creating or editing instant apps uses credits (we generally recommend Expert or higher for this), but using an app after it is created usually uses few or no credits.
Checking your usage
Click the credits icon in the top right to see your available credits at a glance. For more detail:
- Settings → Org → Billing: Your current plan, credit breakdown, and monthly credit allowance.
- Settings → Agent → Usage & Limits: A day-by-day usage chart with a sustainable-per-day line, and per-thread credit breakdown.
- Settings → Org → Usage: Credit usage across all agents and team members in your org.
Controlling your usage
To avoid using more credits than you intend for any given thread, you can set a per-thread daily credit limit in Settings → Agent → Usage & Limits. When a thread hits its daily limit, it pauses until the limit resets at Midnight UTC.
By default, new threads are assigned a limit of 4,000 credits/day, and you can adjust this to any amount you want.
When you run out
Tasklet sends an email at 80%, 95%, and 100% of your credit allowance. When your credits are fully depleted:
- New requests and automations pause until credits are available
- Daily bonus credits let you continue limited usage each day
- You can upgrade your plan at any time for immediate access to more credits
Intelligence Levels
Each Tasklet thread can be set to a different intelligence level. Higher levels are smarter and more capable, but use more credits per action.
| Level | Model | Thinking & Context | Cost | Best for |
|---|---|---|---|---|
| Basic ($) | GPT-5.6 Luna | Low | Lowest | Simple, repetitive tasks like quick lookups, straightforward formatting, and basic triage |
| Advanced ($$) | GPT-5.6 Terra | Moderate | Moderate | Most work. Complex reasoning, multi-step tasks, working across connections, writing, and research |
| Expert ($$$) | GPT-5.6 Sol | Moderate | High | Demanding tasks like instant app creation, detailed analysis across many sources and multi-connection workflows requiring deeper reasoning |
| Genius ($$$$) | Claude Fable 5 | High | Highest | The hardest problems, like long research projects, large-scale processing, and tasks requiring deep sustained reasoning |
We recommend Advanced for most threads as the best balance of capability and cost.
How to change thread intelligence
Select the intelligence level from the dropdown in the thread input. This setting applies to everything that thread does going forward: not just the current conversation, but also all future runs and automations for that thread.
For example, if you set a thread to Advanced and it has a daily email automation, every run of that automation will use GPT-5.6 Terra until you change it. Different threads can use different intelligence levels, so you can run a simple triage thread on Basic while your research thread uses Expert.