How many characters does $10 buy: MAI-Voice-2.1, Flash, Sume TTS

Ten dollars buys about 454,500 characters of Microsoft MAI-Voice-2.1 ($22 per 1M), about 666,600 of MAI-Voice-2.1-Flash ($15 per 1M), and about 210,500 characters on Sume TTS at $0.0475 per 1,000 characters. Microsoft's list is cheaper per character; Sume bundles a request cap, a job result and the rest of its media pipeline.

Where the numbers come from

Microsoft published the two voice prices in the MAI-Transcribe-2-Streaming announcement dated October 1, 2026. Sume's rate is on the rate card and in the catalog; a single Sume TTS request takes up to 20,000 characters.

Sume is the more expensive per character

At list, Sume TTS costs about 2.2 times MAI-Voice-2.1 and about 3.2 times Flash per character. The gap is real; pick on what you need around the audio, and use the Timeline docs if the voice ends up inside a rendered video.

Prorated, not rounded up

The per-character charge is prorated, so a 280-character line is a fraction of a cent rather than a minimum. The catalog rounds shown prices up to whole cents, so use the estimate for quoting a bill and the usage endpoint for what was actually charged.