ElevenLabs text-to-speech models & pricing
9 models listed · appears on 3 of 3 boards · from $40.00/1M
Voice AI company offering text-to-speech, voice cloning, and dubbing APIs across the Eleven model family.
Models on the boards
ModelReleased$/1M charsBoardsComposite
Eleven v4Sept 2026$80.02/384.0
Eleven v3Feb 2026$100.03/378.0
Eleven v4 TurboSept 2026$40.01/371.0
v3 ConversationalAug 2026$50.01/360.0
Turbo v2.5Jul 2024$50.01/343.0
Multilingual v2Aug 2023$100.01/341.0
Flash v2.5Dec 2024$50.02/340.0
Eleven v3 Conversational——1/39.0
Turbo v2——1/37.0
Latency from public benchmarks
Time to first audio, in milliseconds, as each benchmark publishes it. We don't measure latency ourselves, and the sources measure differently (harness, region, sample size), so each figure stays under its own source. Not part of the composite score.
ModelCoval, Time to first audio, p50, us-east-1, millisecondsSpeko, Time to first audio, p50, US East, millisecondsHumanness Index, Time to first audio, median, milliseconds
Eleven v4—844 msmeasured Sep 28, 2026—
Eleven v3——758 msmeasured June 2026
Eleven v4 Turbo192 ms7-day window to Oct 11, 2026205 msmeasured Sep 28, 2026—
v3 Conversational320 ms7-day window to Oct 11, 2026266 msmeasured Jul 3, 2026—
Flash v2.5187 ms7-day window to Oct 11, 2026—197 msmeasured June 2026
Eleven v3 Conversational320 ms7-day window to Oct 11, 2026266 msmeasured Jul 3, 2026—
Turbo v2——302 msmeasured June 2026
Coval: Time to first audio, p50, us-east-1, read Oct 11, 2026 · Speko: Time to first audio, p50, US East, read Oct 11, 2026 · Humanness Index: Time to first audio, median, read Oct 11, 2026. How latency is sourced.