Billing & Pricing

Prismma bills per token — input, output, and (for Claude models) prompt-cache reads/writes each meter separately. There's no seat fee and no monthly minimum: you pay for what you call, deducted from your balance as requests complete.

Prices differ per model and are listed in the model pricing page.

The console displays your balance in EUR/USD. Internally, spend is tracked in raw quota units, where 500,000 quota = $1.00 — you generally never need to think in quota units, but you may see the raw figure in a few advanced usage exports.

For Claude models, prompt-cache reads are billed at a reduced rate versus a fresh input token, and cache writes carry a small surcharge — the same discount/surcharge structure Anthropic uses upstream. Exact per-model cache rates are on the model pricing page.