Plans & capacity
Enterprise & processing capacity
Full-data runs execute on dedicated GPU workers. Every account type comes with a worker allotment — the number of GPUs assigned to your runs — and larger corpora finish proportionally faster with more workers.
Workers by account type
| Account type | Processing workers | What that means |
|---|---|---|
| Academic | 1 worker | One dedicated GPU per run. |
| Individual | 1 worker | One dedicated GPU per run. |
| Organizational | 2 workers | Two GPUs in parallel — large corpora process roughly twice as fast. |
Your run panel shows the workers assigned to each run while it processes. Worker allotments are per account type. Word budgets are unaffected as the same run spends the same words regardless of how many workers process it.
Need it faster?
Organizations with recurring large corpora or deadline-driven workloads can go beyond the standard two workers under an enterprise agreement: custom worker counts, reserved capacity, and priority queueing are all available as separate deals rather than self-service settings.
Higher security & compliance
The self-service platform does not accept regulated data — protected health information, FERPA student records, and similar categories are excluded (see the Regulated Data Notice). If your work requires them, that processing happens under a separate enterprise agreement with the controls it needs: HIPAA processing with a Business Associate Agreement, dedicated single-tenant capacity, or on-premises deployment inside your own infrastructure.
Talk to us
Enterprise agreements are scoped case by case. Write to kai@psyproxy.aiwith your corpus sizes, timelines, and any compliance requirements, and we’ll come back with a concrete proposal.