Eleven Multilingual v2 or Turbo v2.5 on Sume: 28 vs 31 languages
On Sume's TTS Router, Eleven Turbo v2.5 lists 31 languages, 20,000 characters and half the list price of Multilingual v2, which lists 28 languages.
Pick Turbo v2.5 unless you need the Multilingual v2 voice character: in Sume's catalog on main, eleven-turbo-v2.5 lists 31 languages, up to 20,000 characters and a provider list of 50 micro-dollars per character, while eleven-multilingual-v2 lists 28 languages, 10,000 characters and 100 micro-dollars. Turbo v2.5 adds Hungarian, Norwegian and Vietnamese. Both accept stability, similarity boost, style and speed (0.7 to 1.2).
Source: packages/api-contract/src/tts-router-models.ts on origin/main b28f3b0, read 2026-10-13. Sume bills the provider list times 1.25, so these list values become $0.0625 and $0.125 per 1,000 characters before cent rounding.
| Item | Eleven Multilingual v2 | Eleven Turbo v2.5 |
|---|---|---|
| Model id | eleven-multilingual-v2 | eleven-turbo-v2.5 |
| Description in catalog | Lifelike and consistent, 28 languages | Fast and low cost, 31 languages |
| Languages | 28 | 31 |
| Max characters | 10,000 | 20,000 |
| List per character | 100 micro-dollars | 50 micro-dollars |
| Sume per 1,000 characters | $0.125 before cent rounding | $0.0625 before cent rounding |
| Settings | Stability, similarity, style, speed, text normalization | Same |
| Output | mp3_44100_128 | mp3_44100_128 |
When is the second row still right?
When a listening test prefers its voice on your script. The catalog gives no quality ranking, so do a blind test on your own line, with the same voice and language. If the line needs more than 10,000 characters, split it or use Turbo.
What are the newer Eleven rows?
eleven-v4 and eleven-v4-turbo list 80 and 40 micro-dollars with a 10,000 character limit and seed support; eleven-v3 lists 100 with a 5,000 limit. They have no speed setting in the catalog.
Sources
Related posts
More in Models
- Every Sume image model compared: edit, text-only, n and formats
Sume's repo catalog lists 20 image rows: 14 take edit references, 5 are text-only, 1 makes layers. Grok is capped at n 1; Recraft is webp only.
- H3 Max lip sync and audio over 14.8 seconds: split or Fabric
Audio over 14.8 seconds is refused on Sume's H3 Max lip sync because the provider silently clips the rest. Fabric 1.0 takes the long line.
- H3 Max lip sync rejects audio under 5 seconds: what to do
Sume's MiniMax H3 Max lip sync takes 5 to 14.8 seconds of audio and returns invalid_request outside that window. Fabric is the fallback for shorter lines.
- Porting a Sume agent loop to Haiku 5.5: drop temperature, top_p, top_k
Anthropic says to omit temperature, top_p and top_k on Haiku 5.5, and some values return a 400. Strip them from your loop before the first Sume tool call.
Written by Sume