Grok TTS at $15 per million characters vs Sume per-character TTS

xAI's models page lists text to speech at $15.00 per 1M characters, which is $0.015 per 1,000 characters. Sume TTS is also billed per transcript character, at $0.0475 per 1,000 characters on the rate card, with spaces and punctuation counted and a 20000 character maximum per request.

xAI's rate is from its models page; Sume's from the API contract, the pricing FAQ and Generation admission, read 2026-10-01.

How do I convert $15 per million characters?

Divide by 1,000 to get the price per thousand characters: $15.00 / 1,000 = $0.015. A 5,000 character script is then 5 × $0.015 = $0.075 at the xAI list rate. That is plain arithmetic on the page's number; the page does not state per-request examples.

How do I estimate a Sume TTS request?

Count the characters in the transcript, spaces and punctuation included, and multiply by the rate-card price. Sume says paid generation uses public USD estimates, and the pricing FAQ says API calls are billed at each model's published USD rate, so the number you compute is the estimate to compare with the reservation on the usage ledger.

What happens to the estimate when the request runs?

Sume reserves the estimated amount when the request is accepted, captures it on successful completion, and releases or refunds it if the job fails. The ledger shows these as reserved, captured and refunded.

What if my script is over 20000 characters?

Split it into several requests, each at or under the limit. For a model-by-model view of the per-character book, see Cartesia Sonic model ids.