Inworld, the AI research lab and inference provider behind Realtime TTS-2, has acquired Ultravox, the Seattle-based platform developers use to build real-time voice agents; financial terms were not disclosed. All built-in Inworld voices on Ultravox were immediately upgraded to Realtime TTS-2 — a top-ranked model supporting 200+ languages with sub-100ms latency — at no additional cost.
On September 30, 2026, Inworld announced that it has acquired Ultravox, the platform developers use to build real-time voice agents, with the press release distributed via Business Wire on October 1. Financial terms were not disclosed. Ultravox team members have joined Inworld and will continue developing the platform. The two companies had already been working together through a text-to-speech partnership, which the companies say showed how their technologies complement each other and how 'bringing them together could help developers build better voice experiences.'
The acquisition consolidates the voice stack: Ultravox brings expertise in speech understanding, reasoning, and task completion, alongside tools for managing turn-taking and interruptions in live conversations, while Inworld contributes top-of-benchmarks speech models — Realtime TTS-2 and Realtime STT-1 — plus model serving and real-time inference infrastructure for consumer-facing AI applications. CEO Kylan Gibbs framed the deal around continuity and speed: 'Ultravox has built a platform that developers rely on for real-time voice agents... Bringing Realtime TTS-2 to the Inworld voices customers already use is the first benefit we're delivering.' Zach Koch, co-founder and former CEO of Ultravox, added: 'I believe Inworld is the right home for Ultravox and the right team to take it further.'
The first customer-facing change is immediate: all built-in Inworld voices on Ultravox now run on Realtime TTS-2, the top-ranked text-to-speech model on the Artificial Analysis leaderboard, supporting 200+ languages with sub-100ms latency. The upgrade requires no migration or code changes — existing voice IDs continue to work — and comes at no additional cost. TTS-2 also introduces natural-language steering, letting developers control an agent's pace, tone, and emotional delivery through inline instructions, such as asking a voice to slow down when explaining a complicated step or sound warmer when reassuring a customer. Agents using voices from other providers are unchanged 'today,' and existing Ultravox customers keep their accounts with continued support from Inworld. Akhilesh Damaraju, CEO of Equal AI, said his company was already using Inworld's voice technology and called bringing the voices and platform 'under one roof' a natural next step.
Inworld says the acquisition advances its work on speech-to-speech experiences that bring speech understanding, reasoning, and expressive voice generation closer together, with further capabilities and improvements for Ultravox customers promised 'in the coming weeks.' For developers, the practical effect is a single vendor now spanning the model, the inference layer, and the agent platform — reducing the hop count between components in latency-sensitive real-time voice applications.
