ElevenLabsAudio modelActive
Eleven v3 Conversational
PopularityRank 451 of 876
31out of 100same as last month
Summary
Eleven v3 Conversational uses the Text to Dialogue WebSocket for voice-agent dialogue rather than plain TTS. It generates expressive, context-aware speech and supports more than 70 languages. ElevenLabs puts model latency at about 280 ms, excluding application and network time.
Compared with Eleven v3, Eleven v3 Conversational uses the Text to Dialogue WebSocket for the real-time path. ElevenLabs cites about 280 ms of model latency to the response; the figure excludes the agent, network, and application.
All benchmark results
Every published result with its source, test setup and comparison models.
Show all 26 published benchmark results
Page 1 of 3
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.
Gradium hard-case TTS overall sentence pass rate (Show measurement, test conditions, and source)
- Source value
- 65.4
- Score
- 65.4
- Metric
- All-criteria share of sentences where every element is pronounced correctly and completely; higher is better
- Unit
- %
- Benchmark version
- Gradium hard-case TTS set, generated 2026-08; source article 2026-08-31; default settings
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Gradium, Gradium TTS: you no longer have to choose between latency and accuracy
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Vendor-reported result. Gradium hard-case TTS evaluation generated in August 2026 with default settings. The 500-sentence set has 100 sentences per language in English, German, French, Spanish, and Portuguese. Independent native-speaker raters heard randomized, loudness-normalized outputs. The all-criteria sentence pass rate is pooled across ten criteria and averaged across the five languages with equal language weights. Voice IDs and immutable API revisions are not published. Exact table model: ElevenLabs v3 Conversational.
- Context
- Gradium is a competing TTS vendor and reports its own comparison; source type is vendor-reported, not an independent leaderboard. The four comparison models were selected by Gradium. The score is macro-weighted across languages and criteria, so no single pooled sentence denominator or confidence interval is asserted. Voice IDs and serving revisions are unspecified.
Artificial Analysis Provider Voice Arena preference (Show measurement, test conditions, and source)
- Source value
- 1207
- Score
- 1207
- Metric
- blind pairwise native-voice preference Elo
- Unit
- elo
- Benchmark version
- Current Provider Voice Arena board retrieved 2026-10-15
- Category
- audio
- Direction
- Higher is better
- Sample
- 2492
- Confidence interval
- 13
- Lower confidence bound
- 1194
- Upper confidence bound
- 1220
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- superseded
- Retrieved at
- 2026-10-15
- Methodology
- Independent evaluation. Blind pairwise preference on provider-native voice set (8 voices); 2,492 board appearances; model release shown as Aug 2026.
- Context
- Appearances are not unique listeners. The current page does not state a per-model evaluation date range.
Artificial Analysis Provider Voice Arena preference (Show measurement, test conditions, and source)
- Source value
- 1210
- Score
- 1210
- Metric
- blind pairwise preference Elo for provider-native voices
- Unit
- elo
- Benchmark version
- Current Provider Voice Arena board retrieved 2026-10-16
- Category
- audio
- Direction
- Higher is better
- Sample
- 2556
- Confidence interval
- 13
- Lower confidence bound
- 1197
- Upper confidence bound
- 1223
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis
- Status
- superseded
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Exact board row v3 Conversational; blind pairwise preference across eight provider-native voices; 2,556 board appearances. The displayed August 2026 release date is retained from the board.
- Context
- Appearances are not unique listeners. The board does not state a per-model evaluation date range.
Speko TTS human naturalness Elo (English phone-agent lines) (Show measurement, test conditions, and source)
- Source value
- 1590
- Score
- 1590
- Metric
- Blind human A/B naturalness preference Elo; field mean 1500
- Unit
- elo
- Benchmark version
- Speko TTS board; 2026-07-03 board snapshot; captured 2026-10-17
- Category
- text-to-speech / English blind preference
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Speko independent voice benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact Speko internal route elevenlabs:eleven_v3_conversational, displayed model eleven_v3_conversational; English phone-agent lines; naturalness Elo from blind human A/B listening votes, field mean 1500. This is the exact displayed naturalness score, not the separate win-rate field.
- Context
- The evaluator does not publish a per-model sample count, Elo confidence interval, immutable provider checkpoint or full generation controls. The board’s separate 95% interval applies to win percentage against the field, not this Elo value, so it is not copied here. Newer Speko TTS methodology material is not retrofitted to this older board snapshot.
Speko TTS robustness, weighted number/date/currency/operator correctness (Show measurement, test conditions, and source)
- Source value
- 0.3
- Score
- 0.3
- Metric
- Severity-weighted speech-content defect score, 0-1; higher means fewer defects
- Unit
- ratio
- Benchmark version
- Speko TTS robustness board; row measured 2026-07-03; captured 2026-10-17
- Category
- text-to-speech / speech-content robustness
- Direction
- Higher is better
- Source type
- independent-evaluator
- Evaluator
- Speko independent voice benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact Speko internal route elevenlabs:eleven_v3_conversational, displayed model eleven_v3_conversational; robustness quality axis, reported on 0-1. The board defines it as severity-weighted correctness for spoken numbers, dates, currency and operators. It is a content-robustness measure, not naturalness, latency or price.
- Context
- No total test denominator, per-model N, confidence interval, immutable serving revision or complete settings are published. The score is evaluator-defined and task-specific. The S2.1 Pro route is not the distinct Fish Audio S2 Pro profile.
Speko multilingual TTS blind preference Elo (Spanish / es) (Show measurement, test conditions, and source)
- Source value
- 1951
- Score
- 1951
- Metric
- Locale-specific blind human A/B listening preference Elo, es
- Unit
- elo
- Benchmark version
- Speko multilingual TTS board; locale es; measured 2026-07-21; retrieved 2026-10-17
- Category
- text-to-speech / multilingual blind preference
- Direction
- Higher is better
- Lower confidence bound
- 1859
- Upper confidence bound
- 2142
- Source type
- independent-evaluator
- Evaluator
- Speko multilingual TTS benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact Speko internal route elevenlabs:eleven_v3_conversational, displayed model eleven_v3_conversational; language-specific Spanish blind A/B listening study; arena Elo 1951, 95% interval 1859-2142. Each language has its own Elo fit and own field reference; tests use language-matched speech.
- Context
- The locale-specific point estimate and 95% interval are published, but the source gives no per-model observation count, immutable serving checkpoint or complete generation settings. Elo scales are independently fitted and must not be compared across languages.
Speko multilingual TTS blind preference Elo (German / de) (Show measurement, test conditions, and source)
- Source value
- 1798
- Score
- 1798
- Metric
- Locale-specific blind human A/B listening preference Elo, de
- Unit
- elo
- Benchmark version
- Speko multilingual TTS board; locale de; measured 2026-07-21; retrieved 2026-10-17
- Category
- text-to-speech / multilingual blind preference
- Direction
- Higher is better
- Lower confidence bound
- 1724
- Upper confidence bound
- 1879
- Source type
- independent-evaluator
- Evaluator
- Speko multilingual TTS benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact Speko internal route elevenlabs:eleven_v3_conversational, displayed model eleven_v3_conversational; language-specific German blind A/B listening study; arena Elo 1798, 95% interval 1724-1879. Each language has its own Elo fit and own field reference; tests use language-matched speech.
- Context
- The locale-specific point estimate and 95% interval are published, but the source gives no per-model observation count, immutable serving checkpoint or complete generation settings. Elo scales are independently fitted and must not be compared across languages.
Speko multilingual TTS blind preference Elo (French / fr) (Show measurement, test conditions, and source)
- Source value
- 1775
- Score
- 1775
- Metric
- Locale-specific blind human A/B listening preference Elo, fr
- Unit
- elo
- Benchmark version
- Speko multilingual TTS board; locale fr; measured 2026-07-21; retrieved 2026-10-17
- Category
- text-to-speech / multilingual blind preference
- Direction
- Higher is better
- Lower confidence bound
- 1691
- Upper confidence bound
- 1923
- Source type
- independent-evaluator
- Evaluator
- Speko multilingual TTS benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact Speko internal route elevenlabs:eleven_v3_conversational, displayed model eleven_v3_conversational; language-specific French blind A/B listening study; arena Elo 1775, 95% interval 1691-1923. Each language has its own Elo fit and own field reference; tests use language-matched speech.
- Context
- The locale-specific point estimate and 95% interval are published, but the source gives no per-model observation count, immutable serving checkpoint or complete generation settings. Elo scales are independently fitted and must not be compared across languages.
Speko multilingual TTS blind preference Elo (Arabic / ar) (Show measurement, test conditions, and source)
- Source value
- 1564
- Score
- 1564
- Metric
- Locale-specific blind human A/B listening preference Elo, ar
- Unit
- elo
- Benchmark version
- Speko multilingual TTS board; locale ar; measured 2026-07-21; retrieved 2026-10-17
- Category
- text-to-speech / multilingual blind preference
- Direction
- Higher is better
- Lower confidence bound
- 1511
- Upper confidence bound
- 1621
- Source type
- independent-evaluator
- Evaluator
- Speko multilingual TTS benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact Speko internal route elevenlabs:eleven_v3_conversational, displayed model eleven_v3_conversational; language-specific Arabic blind A/B listening study; arena Elo 1564, 95% interval 1511-1621. Each language has its own Elo fit and own field reference; tests use language-matched speech.
- Context
- The locale-specific point estimate and 95% interval are published, but the source gives no per-model observation count, immutable serving checkpoint or complete generation settings. Elo scales are independently fitted and must not be compared across languages.
Speko multilingual TTS blind preference Elo (Filipino / fil) (Show measurement, test conditions, and source)
- Source value
- 1693
- Score
- 1693
- Metric
- Locale-specific blind human A/B listening preference Elo, fil
- Unit
- elo
- Benchmark version
- Speko multilingual TTS board; locale fil; measured 2026-07-21; retrieved 2026-10-17
- Category
- text-to-speech / multilingual blind preference
- Direction
- Higher is better
- Lower confidence bound
- 1597
- Upper confidence bound
- 1815
- Source type
- independent-evaluator
- Evaluator
- Speko multilingual TTS benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact Speko internal route elevenlabs:eleven_v3_conversational, displayed model eleven_v3_conversational; language-specific Filipino blind A/B listening study; arena Elo 1693, 95% interval 1597-1815. Each language has its own Elo fit and own field reference; tests use language-matched speech.
- Context
- The locale-specific point estimate and 95% interval are published, but the source gives no per-model observation count, immutable serving checkpoint or complete generation settings. Elo scales are independently fitted and must not be compared across languages.
Speko multilingual TTS blind preference Elo (Norwegian / nb) (Show measurement, test conditions, and source)
- Source value
- 1628
- Score
- 1628
- Metric
- Locale-specific blind human A/B listening preference Elo, nb
- Unit
- elo
- Benchmark version
- Speko multilingual TTS board; locale nb; measured 2026-07-21; retrieved 2026-10-17
- Category
- text-to-speech / multilingual blind preference
- Direction
- Higher is better
- Lower confidence bound
- 1533
- Upper confidence bound
- 1733
- Source type
- independent-evaluator
- Evaluator
- Speko multilingual TTS benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact Speko internal route elevenlabs:eleven_v3_conversational, displayed model eleven_v3_conversational; language-specific Norwegian blind A/B listening study; arena Elo 1628, 95% interval 1533-1733. Each language has its own Elo fit and own field reference; tests use language-matched speech.
- Context
- The locale-specific point estimate and 95% interval are published, but the source gives no per-model observation count, immutable serving checkpoint or complete generation settings. Elo scales are independently fitted and must not be compared across languages.
Speko multilingual TTS blind preference Elo (Hindi / hi) (Show measurement, test conditions, and source)
- Source value
- 1490
- Score
- 1490
- Metric
- Locale-specific blind human A/B listening preference Elo, hi
- Unit
- elo
- Benchmark version
- Speko multilingual TTS board; locale hi; measured 2026-08-05; retrieved 2026-10-17
- Category
- text-to-speech / multilingual blind preference
- Direction
- Higher is better
- Lower confidence bound
- 1362
- Upper confidence bound
- 1600
- Source type
- independent-evaluator
- Evaluator
- Speko multilingual TTS benchmark
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. Exact Speko internal route elevenlabs:eleven_v3_conversational, displayed model eleven_v3_conversational; language-specific Hindi blind A/B listening study; arena Elo 1490, 95% interval 1362-1600. Each language has its own Elo fit and own field reference; tests use language-matched speech.
- Context
- The locale-specific point estimate and 95% interval are published, but the source gives no per-model observation count, immutable serving checkpoint or complete generation settings. Elo scales are independently fitted and must not be compared across languages.
Page 1 of 3
Popularity compared
How often the model is searched for, used and mentioned, compared with every model of its category.
Popularity index
0 to 100, higher is more popular
20 of 83 models
As of November 2026 · Source: gradually.ai popularity index
Over time
Weekly values from October 18, 2026 to November 6, 2026
Eleven v3 ConversationalGPT-Transcribe 32Index-Echo-S2ST-2B 30Voxtral Mini TTS 29MOSS-Audio-4B-Thinking 29
Method changes in the timeline, a jump at these points is no market event:
- 1November 1, 2026: Value only with search or usage data, 30-day attention
What is gaining popularity right now, across models, vendors, harnesses and AI subscriptions, is collected in All AI trends.
Eleven v3 Conversational compared
Similar models
More models of the same kind as Eleven v3 Conversational, chosen by model line, developer, size, price and release date.
- Eleven v3ElevenLabs · same developer
- Scribe v2ElevenLabs · same developer
- Scribe v2 RealtimeElevenLabs · same developer
- Eleven Flash v2.5ElevenLabs · same developer
- Eleven Multilingual STS v2ElevenLabs · same developer
- Eleven Text to Sound v2ElevenLabs · same developer
- Eleven Music v2.5ElevenLabs · same developer
- Scribe v2 MedicalElevenLabs · same developer
All specifications
Architecture
Input and output
Watermarking and provenance
AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.
- AI watermarking
- Subject to limitations. ElevenLabs describes an inaudible watermark in its own generated audio and identifies SynthID, including for speech generation. Older recordings made before its introduction and third-party model outputs are not automatically covered.
- Source 1: AI watermarkingSource 2: AI watermarkingSource 3: AI watermarkingSource 4: AI watermarking
- Watermarking technique
- SynthID
- Source 1: Watermarking techniqueSource 2: Watermarking techniqueSource 3: Watermarking techniqueSource 4: Watermarking technique
- Labelling reviewed
- Nov 5, 2026
- Source 1: Labelling reviewedSource 2: Labelling reviewedSource 3: Labelling reviewedSource 4: Labelling reviewed
Licence and access
API
Further details
Sources and data date
Show all 32 sources
Benchmark sources at a glance
Which independent evaluators tested this model. This is coverage, not a quality ranking.
No active result from the reviewed external evaluators can currently be attributed to this model. Other source evidence remains available below.
Without vendor claims and historical or disputed results; those stay labelled in the single values.