ElevenLabs launches Eleven v4 and v4 Turbo voice models
ElevenLabs split its new v4 release between higher-control production speech and low-latency agent audio.
Eleven v4 supports 90-plus languages, multiple speakers, sound effects, emotional direction tags, and up to 10,000 characters per generation. Eleven v4 Turbo is built for live voice agents, with bidirectional streaming and claimed median timings of 100 ms inference latency and 150 ms to first speech. Both models are available through ElevenLabs products and the ElevenAPI, and Turbo is also available through ElevenAgents. Older cloned voices may need retraining, while Professional Voice Clones return after being unavailable in v3. TestingCatalog's note
Eleven v4 supports 90-plus languages, multiple speakers, sound effects, emotional direction tags, and up to 10,000 characters per generation. Eleven v4 Turbo is built for live voice agents, with bidirectional streaming and claimed median timings of 100 ms inference latency and 150 ms to first speech. Both models are available through ElevenLabs products and the ElevenAPI, and Turbo is also available through ElevenAgents. Older cloned voices may need retraining, while Professional Voice Clones return after being unavailable in v3. TestingCatalog's note
score 6