Microsoft MAI-Voice-2.1 costs $22 per 1M characters; Flash is $15

Microsoft's MAI-Voice page lists MAI-Voice-2.1 at $22 per 1M characters and MAI-Voice-2.1-Flash at $15 per 1M characters. A 5,000-character script costs $0.1100 and $0.0750. Sume TTS 1.0 is $0.0475 per 1,000 characters, which is $0.2375 for the same script.

Microsoft's figures are from its MAI-Voice page, read 2026-10-01; Sume's from the rate card.

What does the script math look like?

Price per million characters times 0.005 for a 5,000-character script.

Why is Flash cheaper?

The same page lists a much lower latency figure for Flash (about 45 ms against about 550 ms). It does not say why Flash is cheaper or that it is a smaller model, so do not read the price gap as a quality ranking.

Is Sume's rate comparable?

It is a per-character rate on the same unit, but Sume TTS 1.0 has no engine picker: model and model_id are rejected with a 400 on TTS 1.0, and the TTS Router catalog is Sonic only, per the API reference. So Microsoft's voices are not selectable on Sume; this is a price reference, not a swap.

What should I measure before choosing?

Price per finished minute, not per character: a voice that speaks slower uses more seconds of audio for the same text, and you pay per character either way. Time a real script on each, then compare listening quality at the same speed.