Currency and Billing
The Shared IB-X Currency Pool
The IB-X Currency SKU provides a single currency pool associated with your subscription. Supported Gateway services consume this pool according to their usage units and applicable rates.
Purchase and associate currency with the subscription used by your automation. Without an active currency allocation, Integration Gateway services are not available.
Current Rates
Consumption rates can change. These pages explain usage units and metering without publishing fixed rates. Confirm the current rate for the service, activity, and model when purchasing or topping up IB-X Currency through your IntelliBuddies purchasing contact.
IB-X Currency is the unit consumed by the gateway. A service's currency consumption rate and the monetary price of purchasing currency are separate quantities.
How Usage Is Measured
| Service | Usage unit | What to count |
|---|---|---|
| Azure Document Analysis | Page | Pages in the submitted document |
| Azure Cognitive Text | Text record of 1,000 characters | Input text rounded up to a whole text record |
| OpenAI Gen-AI | Token | Input, cached input, and output usage at their respective model rates |
| Llama Gen-AI | Block of 1,000 tokens | Combined input and output tokens rounded up to a whole block |
| IntelliBuddies Cognitive Services | Request | Requests for the selected activity |
| IntelliBuddies Document Analysis | Call | Calls to the selected activity |
| Deepgram Speech-to-Text | Second | Audio duration for the selected model |
| Deepgram Text-to-Speech | 1,000 characters | Input text for synthesis |
| Deepgram Add-on | Second | Audio duration for each enabled feature |
| Sarvam Speech-to-Text | Second | Audio duration for transcription or translation |
| Sarvam Text-to-Speech | Character | Input text for synthesis |
| Sarvam Language Model | Token | Input, cached input, and output usage |
For one usage category, multiply billable units by the applicable rate. For multiple categories, calculate each separately and add the amounts.
Examples Without Fixed Rates
- A two-page document represents two page units for Azure Document Analysis.
- An Azure text input of 2,500 characters represents three text records.
- A Llama request using 9,500 combined input and output tokens represents ten blocks of 1,000 tokens.
- Ten calls to Invoice Analyzer represent ten call units. The activity supports single-page invoices only.
Do not apply one service's rounding rule to another. These pages do not specify OpenAI token rounding or how rounding is applied across multiple requests. Confirm these details with your IntelliBuddies contact when estimating a workload.
Language Model Token Categories
Input includes the prompt, system instructions, and conversation history. Cached input is the reusable portion reported as cached by the service; output is the generated response. Cache use depends on the model and service and should not be assumed for every request.
When usage reports total input inclusive of cached tokens, subtract cached tokens to obtain uncached input before applying the ordinary input rate. Count each input token once. Use the actual usage categories reported by the service and the applicable model rates.
Plan and Verify Consumption
Estimate each activity separately, including repeated calls, document sizes, and expected generated output. Add the amounts to estimate consumption from the shared pool. Chat history can increase input token usage on later requests.
Compare estimates with actual consumption after running a representative workload. For Enterprise subscriptions, the Integration Gateway Credit Summary provides available, consumed, reserved, and total credits plus consumption history. See Usage and Troubleshooting.