Audio modelActive
Eleven v4
ElevenLabs
- Released
- September 28, 2026
- Data date
- October 7, 2026
- Model class
- Languages
Eleven v4 generates speech that adapts tone, pacing, and emotion to the text's context. ElevenLabs recommends it for high-fidelity content such as audiobooks and narration, and lists more than 90 supported languages.
For real-time applications, ElevenLabs offers Eleven v4 Turbo as the low-latency variant.
Specifications and access
| Specification | Value and source |
|---|---|
| Model class | Text-to-speech modelSource |
| Access | ElevenAPI and ElevenLabs productsSource |
| Input modalities | TextSource |
| Output modalities | AudioSource |
| API model ID | eleven_v4Source |
| Languages | More than 90Source |
| Focus | Rhythm, emotion, pacing, character voice, and conversational context; recommended for high-fidelity content such as audiobooks and narrationSource |
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
ElevenLabs blind pairwise listener preference (Show measurement, test conditions, and source)
- Source value
- 75
- Score
- 75
- Metric
- approximate listener preference share
- Unit
- %
- Benchmark version
- ElevenLabs comparison reported September 2026
- Category
- audio
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- ElevenLabs Eleven v4 launch post
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Vendor-reported result. Blind pairwise setup used the same line; listeners judged expressiveness and naturalness. Named competitors: Cartesia Sonic 3.6, Inworld TTS-2, Gemini 3.8 Flash-Lite, and Gemini 3.8 Flash. The source publishes neither sample count nor weighting.
- Context
- Vendor-run comparison without published sample count, per-competitor results, or independent replication. The result is explicitly approximate.
Artificial Analysis Provider Voice Arena Elo (Show measurement, test conditions, and source)
- Source value
- 1321
- Score
- 1321
- Metric
- human blind-pairwise voice preference Elo
- Unit
- elo
- Benchmark version
- September 2026 snapshot
- Category
- audio
- Direction
- Higher is better
- Sample
- 1930
- Confidence interval
- 18
- Lower confidence bound
- 1303
- Upper confidence bound
- 1339
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis Provider Voice Arena
- Status
- active
- Retrieved at
- 2026-10-07
- Methodology
- Independent evaluation. Blind pairwise evaluation used each provider's native voice set; 1,930 model appearances across eight voices. Higher Elo means more frequent preference. September 2026 snapshot.
- Context
- Appearances are not a count of unique people or prompts. The snapshot does not give a per-model evaluation date range.
Sources and data date
| Type | Evidence and data date |
|---|---|
| Research status | Research date October 7, 2026. 2 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not published | blind-test-sample-size Original wording ElevenLabs does not publish listener counts or per-competitor results for its approximate preference share. |
| Additional source | ElevenLabs Eleven v4 announcement (retrieved October 7, 2026) · ElevenLabs Eleven v4 announcement · Editorial description reviewed October 7, 2026 |
| Additional source | ElevenLabs model documentation (retrieved October 7, 2026) · ElevenLabs model documentation · Editorial description reviewed October 7, 2026 |
| Additional source | Artificial Analysis Provider Voice Arena (retrieved October 7, 2026) |