ElevenLabs has unveiled a new AI model it says outperforms rivals built by tech titans Google and OpenAI, as the startup works to cement its status as a top industry player following a $180m raise in January.
The New York and London-based company announced on Wednesday it had released its first speech-to-text model, having made a name for itself in the audio-generation space.
The model — which converts spoken language into written text — produced a greater accuracy rate in benchmark tests across 99 languages than competitors including Google’s Gemini 2.0 Flash and OpenAI’s Whisper Large V3, ElevenLabs says.
The launch comes a month on from the startup announcing a $180m Series C, led by US investors Andreessen Horowitz and Iconiq Growth, which took its valuation to $3.3bn.
Founded in 2022 by Google and Palantir alumni, ElevenLabs is building AI audio tech and has previously focused on synthetic voice generation — converting text to speech and allowing publishers and content creators to turn written material into audio.
Its new speech-to-text model can transcribe speech in 99 languages, the company says, and signals a move towards audio transcription.
A competitive space, leaders in the field include US-based Otter and Sweden’s Sana.ai, as well as numerous Big Tech players.
“Our largest potential competition is OpenAI,” founder Mati Staniszewski told Sifted in July, citing research capabilities, huge resources and vastly more funding. “They will probably start to build more and more models in audio.”
OpenAI integrated voice responses on ChatGPT last year. Google announced an update to its Gemini 2.0 speech-to-text model in December, in a bid to beef up capabilities of its AI chat function.
ElevenLabs currently has more than 200 employees, according to LinkedIn. Its product and research teams are primarily based in Europe, in the UK, alongside Poland and Hungary, while its US footprint is mostly made up of sales and partnerships teams.
Read the orginal article: https://sifted.eu/articles/elevenlabs-new-ai-model-outperforms-google-openai-news/