We’re building for developers scaling voice AI. Introducing two features to make collaboration and visibility easier: 🤝 Organizations and 📊 Dashboards. 🤝 The Organizations feature gives teams shared access to API keys, custom voices, and billing–all under one account.

Headed to Customer Contact Week in Vegas next week? Come find us on the floor! We’re building the next generation of real-time Voice AI–faster, more flexible, and ready for the enterprise–and we’d love to meet you. Swing by our booth to see what we’ve been working on, chat with two of

When we put lots of text (eg a code repo) into LLM context, cost soars b/c of the KV cache’s size. What if we trained a smaller KV cache for our documents offline? Using a test-time training recipe we call self-study, we find that this can reduce cache memory on avg 39x

Building voice agents? Meet Ink-Whisper: the fastest, most affordable streaming speech-to-text model. 🌎 Optimized for accuracy in real-world conditions 👯 Pair with our Sonic text-to-speech → fastest duo in voice AI 🔌 Plugs into Vapi,Pipecat AI, LiveKit Read more:

Talk to Cartesia speech-to-text about Cartesia speech-to-text. Cartesia launched a streaming STT model today, called Ink-Whisper, that's optimized for realtime voice AI. Pipecat AI has launch-day support for this new model, so I figured I'd talk to the model about itself.

Ink-Whisper is fast! Cartesia released a new STT model yesterday, and it's as fast as you'd expect. Streamed transcription finishes in <100ms—before you can say the next sentence. I used it to build a live teleprompter that follows along with what you're saying. Code in 🧵

👑 We’re #1! Sonic-2 leads @Labelbox’s Speech Generation Leaderboard topping out in speech quality, word error rate, and naturalness. Build your real-time voice apps with the 🥇 best voice AI model. ➡️ labelbox.com/leaderboards/s…

Thanks to Sapphire Ventures for hosting our co-founder Brandon Yang at the Hypergrowth Engineering Summit. He shared that voice is the next UX frontier and building voice agents requires excellence at every layer. Great to see so many technical leaders leaned in! #SapphireHypergrowthEng

Despite theoretically handling long contexts, existing recurrent models still fall short: they may fail to generalize past the training length. We show a simple and general fix which enables length generalization in up to 256k sequences, with no need to change the architectures!

I converted one of my favorite talks I've given over the past year into a blog post. "On the Tradeoffs of SSMs and Transformers" (or: tokens are bullshit) In a few days, we'll release what I believe is the next major advance for architectures.

Tokenization has been the final barrier to truly end-to-end language models. We developed the H-Net: a hierarchical network that replaces tokenization with a dynamic chunking process directly inside the model, automatically discovering and operating over meaningful units of data

We're excited to announce a new research release from the Cartesia team, as part of a long-term collaboration to advance deep learning architectures. We've always believed that model architectures remain a fundamental bottleneck in building truly intelligent systems. H-Nets are

The micro1 research team analyzed more than 300,000 AI-led interviews to assess the performance of different combinations of speech-to-text engines, large language models, and text-to-speech systems, now including Cartesia’s latest TTS engine. Each stack was evaluated for

We designed Line with everything we learned about how best in class AI agents are built. - Use code. The best AI experiences are too complex to be built any other way. - Build iteratively with evals. Ultimately evals define the capabilities of any AI product (agents or models) -

Sales reps are constantly on the move. How can they get critical deal insights without breaking their stride? The answer is voice AI. We've partnered with Rox to bring our real-time voice AI to their Command AI chat assistant. With it, reps can get pre-meeting briefs and

Earlier this week, we hosted our YC Voice Agents event with an incredible lineup of panelists: Hassaan Raza at Tavus (YC21) Anthony Krivonos at Toma (YCW24) Arkadiy Telegin at Leaping AI (YC W25) (YCW25) Max Child 🌐 at Volley (YC18) YC companies were among Sonic’s earliest

Build a real-time AI Voice Agent that can actually sell. In less than 30 minutes, you'll learn how to build a sophisticated real-time voice sales agent that can have natural conversations with potential customers. Includes a full code notebook, recorded tutorial, and docs The

I'll be speaking on a panel on Frontier Speech Models at Vapicon and will be hanging out this afternoon! Come say hi!