Your own API key (BYOK): you enter your OpenAI, Azure, Gemini or DeepSeek key in the program options and pay the provider directly. ChronoScan is not involved in the cost.
ChronoScan credits: you buy credit packs in the ChronoScan HUB and ChronoScan handles the provider. No accounts or cards with each provider; the consumption is visible in the HUB per installation, job and model.
Credits are also consumed by the cloud OCR engines (Google Vision, per page) when you do not use your own key. Local models (Ollama, local OpenAI-compatible server) consume nothing.
Estimating before processing
The AI Cost Estimate button (AI panel of the Capture tab) calculates, for the open batch, how much the processing will cost per model and per mode (synchronous or deferred) and compares models, so you can choose the cheapest model that gives good results. With Flash or mini class models an invoice costs about a euro cent or less; the deferred mode halves it.
Watching the consumption
ChronoScan Cr. (Help panel): balance and information about your credits.
Reports (Help panel): activity reports with batches, documents, pages, errors, exports and credits consumed; the "Local AI servers" tab keeps the history of the requests that did not go through the HUB.
The program warns when the balance is low. Without balance the AI and cloud OCR features stop; everything else keeps working.
Each document keeps the token usage of its requests in the raw response files (oai.<alias>.json, gemini.<alias>.json); Gemini also bills the thinking tokens. To compare engines or prompts, duplicate the same batch and compare the results document by document.
