Microsoft MAI-Voice-2.1 at $22 per 1M characters vs ElevenLabs v4
MAI-Voice-2.1 lists at $22 per million characters, equal to ElevenLabs v4's sale rate until Oct 12 and well under its $80 list. Sume TTS is $47.50 per million.
Microsoft's MAI-Voice-2.1 is listed at $22 per million characters, the same $0.022 per 1K that ElevenLabs v4 shows on sale until Oct 12; after the sale v4 returns to $80 per million, 3.6x the Microsoft price. Sume TTS is $47.50 per million. The Flash variant of MAI-Voice is listed at $15 per million.
The Microsoft figures come from its Foundry announcement; I could only confirm them from a search summary of that post, so check Azure's own pricing page for the final number. The ElevenLabs figures are from its API pricing page (read 2026-10-04).
| Model | Per 1K characters | Per 1M characters |
|---|---|---|
| Microsoft MAI-Voice-Flash | $0.015 | $15.00 |
| Microsoft MAI-Voice-2.1 | $0.022 | $22.00 |
| ElevenLabs v4 (sale until Oct 12) | $0.022 | $22.00 |
| Sume TTS | $0.0475 | $47.50 |
| ElevenLabs v4 (list) | $0.08 | $80.00 |
The ratios
MAI-Voice-2.1 at $22 against ElevenLabs v4 list at $80 is 80 / 22 = 3.64x. Against Sume at $47.50 it is 47.5 / 22 = 2.16x. Flash at $15 is 1.47x cheaper than 2.1 (22 / 15) and 3.17x cheaper than Sume (47.5 / 15).
On a 100,000-character script the totals are $2.20 for MAI-Voice-2.1, $4.75 for Sume and $8.00 for ElevenLabs v4 at list; the sale row is also $2.20.
Where Sume fits
Sume's TTS is billed per transcript character at $0.0475 per 1K, with spaces and punctuation counting and a 20,000-character job limit. It sits inside the same account as video, image and timeline jobs, which is the reason to choose it: the narration, the clip and the final render share one balance and one set of refunds. If you only buy speech in bulk and have a Microsoft agreement, the Microsoft per-character rate is lower.
- Pure bulk narration: compare MAI-Voice and the ElevenLabs sale.
- Narration inside a video pipeline: Sume keeps the whole chain under one API.
Steps
- Open the Azure pricing page for your region and confirm the MAI-Voice rate.
- Count your characters, divide by 1,000, multiply by each rate.
- Audition the same 300 characters on each voice.
- Check the Oct 12 sale end date for ElevenLabs if your run is long.
A 2-million-character month
At two million characters a month, MAI-Voice-2.1 is $44, MAI-Voice-Flash is $30, Sume is 2 x $47.50 = $95 and ElevenLabs v4 at list is $160. The sale row is $44 for as long as it lasts, then it moves to the $160 row.
Caveats
Regional pricing, commitment tiers and voice availability can change Microsoft's figure. This page compares list per-character rates only, not quality, latency or licensing terms.
Sources
Related posts
More in Comparisons
- Midjourney has no public API: comparable control in a pipeline
Midjourney's 10/1 alpha adds a pinnable --exp setting and edits that keep aspect ratio. What to use when you need that kind of control from code.
- Midjourney moodboards now take uploads: reference sets for Sume edits
Midjourney's October 1 changelog lets moodboard pages accept file uploads. To reuse a reference set in an API, store the URLs and send them as input_references.
- MiniMax H3 768p 'half the price' claim checked against Sume rows
MiniMax says H3 at 768p costs under half of mainstream 720p. On Sume's own rows H3 768p is $0.075 per second against $0.125 to $0.14 at 720p. The numbers.
- MiniMax H3 vs H3-Max limits: the MiniMax page next to Sume's catalog
MiniMax lists H3 at 4 to 15 s and H3-Max at 5 to 15 s with different resolutions. Sume's catalog states its own ranges. Side by side, with the file-size limits.
Written by Sume