Credits & Billing

How AI Credits Work

Your organization's AI credit balance is granted with your subscription — there is no separate credit purchase flow.

A free organization receives 100 credits once, when it is created. That grant is not renewed: once it is spent, AI features stop until the organization upgrades.

A paid organization receives 100 credits per seat, every month, at the same rate on Premium and Enterprise, so seat count is the only thing that changes it.

• On a paid plan the balance is overwritten on the 1st of every calendar month with a fresh allocation.
• Unused credits are not carried over — the balance resets, it does not accumulate.
• The free grant has no reset. Cancelling a paid plan returns the organization to free and clears the balance; it does not issue a second grant.
• Credits are not individually purchasable as top-ups, and upgrading does not raise the per-seat rate; to change a paid allocation, change your seat count.
• Only credit-consuming AI features (Test Case Generation and the AI Agent) draw down the balance, and they are charged token-based on actual usage. Insight reports and the Flakiness/Duplicate score badges consume no credits.

What an AI Action Costs

There is no flat per-action price. Every charge is calculated the same way:

credits = (tokens sent + tokens returned) ÷ the tokens-per-credit rate for that feature's quality level

Tokens sent includes your prompt plus everything TestKase adds around it — instructions, and for the AI Agent the conversation history and project context. Tokens returned is the model's reply. The rate depends on the quality level you have set for that feature on the Organization → AI Credits page, and the two features are set independently:

Read the table as "how much work one credit buys". The same request costs roughly 12 times more on Premium than on Basic for the AI Agent, and roughly 8 times more for test case generation. Both features start on Balanced.

Budgeting a single action

The size of one action is bounded, so you can put a ceiling on it:

• One generation request returns at most 16,384 tokens, whether you asked for 1 test case or 10. Output alone therefore tops out at roughly 0.7 credits on Basic, 5.5 on Balanced and 5.7 on Premium — plus whatever your requirement text costs to send.
• One AI Agent message gets a reply of at most 4,096 tokens per model call, and the agent may make up to 10 tool-calling rounds before it answers you. A simple question is one round; a request that searches, creates and then verifies is several, and each round re-sends the conversation so far. This is why long sessions get more expensive per message.
• Generating 10 test cases in one request costs far less than 10 separate requests, because the instructions and your requirement are sent once instead of ten times.

When the balance changes

• Credits are deducted after a usable answer comes back, in the same transaction that writes the row you see in Credit Usage History. A request that errors — no credits, AI disabled, or a reply TestKase cannot read — is not charged, even if the model did some work.
• The balance is checked before the call and only refuses when it has reached zero. A request that starts with 0.5 credits left runs to completion even if it costs 5 — so a single expensive action can take the balance below zero. The next action is then refused.
• Charges are fractional and stored to two decimal places. A short agent message on Basic can round to 0.00 credits.
• When the balance is exhausted, both test case generation and the AI Agent refuse with a message that depends on the plan — a paid organization is told "You've used this month's AI credits. They refresh on the 1st, or add seats to raise your monthly allowance.", and a free organization is told "You've used your free AI credits. To use additional AI credits, upgrade to a Premium plan." There is no top-up. The balance is restored by the monthly reset — or immediately if you activate or change a paid TMT subscription, which overwrites the balance with 100 × your new seat count on the spot rather than waiting for the 1st. Note that this is an overwrite in both directions: reducing seats lowers the balance immediately too.

Credit Management

Effective credit management ensures your team always has AI capabilities when needed:

Monitoring Usage

• Balance indicator — The credit balance is shown in the navigation bar when AI is enabled.
• Usage breakdown — The Organization → AI Credits page lists every AI call: date, user, which feature, and credits consumed.
• Monthly reset (paid plans) — On a paid plan the balance is replenished on the 1st of each month, and usage in the current month draws down that month's allocation. A free organization has no reset — its 100-credit grant is all it gets.

Controlling Costs

• Check the AI quality level — The Basic / Balanced / Premium setting on the AI Credits page is the single biggest lever on cost: roughly 12x for the AI Agent and 8x for test case generation between the cheapest and most expensive level. See What an AI Action Costs and AI Privilege & Quality Settings.
• Disable AI temporarily — Toggle AI off during periods when it is not needed (e.g., between sprints). AI is on by default, so this is an opt-out, not an opt-in.
• Educate your team — Share guidelines on when AI features are most valuable to avoid wasteful usage.
• Keep AI Agent sessions focused — Long conversations carry more context and therefore use more tokens per message; start a fresh session for unrelated tasks.
• Refine generation prompts — Better descriptions produce better results on the first try, reducing the need for multiple generation attempts.

Data Privacy & Security

TestKase takes data privacy seriously when processing AI requests:

• Encrypted in transit — All data sent to AI models uses TLS encryption.
• Not used for training — Under our providers' API terms, data sent through the API is not used to train their models. TestKase does not send your data to any provider for training or analytics.
• Limited provider-side retention — Providers may retain API request content for a short period for abuse monitoring, under their own terms.
• Organization-scoped — AI features only access data within your organization. There is no cross-organization data sharing.
• Audit trail — Every AI call is recorded in Credit Usage History on the Organization → AI Credits page, with the date, the user, the feature and the credits consumed.