June 16, 2026

Cartesia unveils Sonic 3.5 and Ink 2, top-ranked voice models

Cartesia has announced the release of Sonic 3.5 and Ink 2, its new models that function as a unified real-time voice stack for transcription and speech synthesis. With Ink 2 ranking first for accuracy in streaming speech-to-text, and Sonic 3.5 leading in real-time text-to-speech with a response time of around 82ms, Cartesia solidifies its position as the only provider to offer top-ranked models for both functionalities in the market. These advancements stem from innovative architectures that enhance both speed and quality, making the models suitable for integration in voice agents.

Ink 2: Ink 2 is a streaming speech-to-text model from Cartesia with integrated capabilities for turn detection and noise handling. It is designed for English-language use with robustness to accents and aims to simplify voice agent pipelines by reducing the need for auxiliary components. The model ranks at the top of relevant benchmarks and pairs with Sonic 3.5 in the release.
Cartesia: Cartesia is an AI company specializing in real-time voice technology and models for transcription and speech synthesis. The company develops state-of-the-art architectures optimized for low-latency performance in voice agents and similar applications. It is directly responsible for releasing the Sonic 3.5 and Ink 2 models highlighted in the news.
Krandiash: Krandiash is the individual behind the official announcement of Cartesia’s latest models via social media. The post emphasizes the company’s unique position as the provider of top-ranked models for both speaking and listening in real-time voice agents. This directly ties the entity to the news event as the quoted source of the release.
Sonic 3.5: Sonic 3.5 is a real-time text-to-speech model developed by Cartesia that supports multilingual capabilities and features such as voice cloning and handling of context-specific pronunciation. It is positioned as a leading option for low-latency speech output in agentic systems. The model forms one half of the single-stack voice solution announced in the news.

Market Position: Cartesia is the only provider offering the top-ranked models simultaneously for both text-to-speech and speech-to-text in streaming scenarios.
Product Integration: The models are built as a unified real-time voice stack that combines transcription and speech synthesis for direct use in voice agents.
Architecture Innovation: New model architectures from Cartesia enable improved frontiers in both speed and quality for streaming voice tasks.

AI tools built by Emerald Force

Built and supported by Emerald Force.

You might also like