According to CoinMeta, ElevenLabs has released a new generation of voice models: v4 and the real-time version v4. turbo features more natural emotion and voice control, while turbo is designed for real-time voice interactions with an average inference delay of about 100ms. In the latest blind test rankings artificial, analysis, and TTS, Eleven and v4 ranked first with 1315 points, surpassing Cartesia Sonic with 3.6 points, Gemini with 3.8 points, Flash TTS with 1267 points, and Qwen-Audio-3.0-TTS-Plus with 1258 points. The previous generation, Eleven and v3, currently only has 1169 points and ranks 17th. v4 has also seen an expansion in language coverage, increasing from over 70 to more than 90 languages, allowing users to directly specify tone, rhythm, emotion, and sound effects.