A real-time voice model and API that handles interruptions, adapts mid-sentence, and carries conversations the way people actually talk. It leads major deployed real-time APIs on audio benchmarks, supports 100+ languages, and is compatible with the OpenAI Realtime API.

A benchmark for post-interruption recovery in voice agents — whether what the model says after you cut in resumes the workflow correctly. 428 interruption samples, 6 types, 10 enterprise domains, closed- and open-weight models.

Developers can now generate talking-head avatar videos from a still image and either an audio clip or Higgs TTS text input.

Higgs TTS 3 is built for voice chat: it speaks, not just reads. It turns model responses into expressive conversational speech across 100 languages, with zero-shot voice cloning and inline control over emotion, style, prosody, pauses, and sound effects.

A benchmark for conversational proactivity in LLMs — noticing and acting on what the user implied but never said. 198 curated dialogues, 624 trigger points, 16 models, and a leaderboard where Recovery proves dramatically hard.

A real-time foundation model that brings human-like digital presence to customer conversations, virtual assistants, training, and interactive experiences