Inworld AI is a research lab focused on realtime voice AI, providing the voice layer for consumer AI applications. Developers use Inworld to give their applications natural, low-latency voice conversations at scale. The platform includes Realtime TTS-2, a text-to-speech model supporting more than 100 languages with natural-language voice steering, alongside speech-to-text, an LLM Router with access to more than 200 models at provider rates, dedicated GPU infrastructure, and a Realtime API that unifies speech-to-text, language models, and text-to-speech into a single live conversation. Each product is available through its own API, with unified billing. Inworld is built for realtime conversation at scale: roughly 200-millisecond responses at high concurrency, with cost efficiency that holds as usage grows. Its text-to-speech models are among the top-ranked on the independent Artificial Analysis Speech Arena. Because Inworld ships models and APIs rather than consumer products of its own, it does not compete with the applications built on it. Inworld powers voice applications across companionship, education, coaching, and entertainment, with customers including Status by Wishroll, Bible Chat, Talkpal, NVIDIA, NBCUniversal, and Logitech Streamlabs. The founding team led product for LLMs at DeepMind and built Dialogflow, the conversational AI platform acquired by Google. Founded in 2021 and headquartered in Mountain View, California, Inworld has raised more than $125M from investors including Lightspeed, Kleiner Perkins, Founders Fund, CRV, Microsoft's M12, Meta, Intel Capital, Samsung NEXT, LG Technology Ventures, and Bitkraft.