KG
Karan Goel
Co-founder & CEO
Previously PhD researcher at Stanford University
Stanford (PhD)
Stanford ML researcher; co-author of foundational state-space-model work. Co-founded Cartesia in 2023 and serves as CEO.
Cartesia builds real-time multimodal AI models powered by state-space model (SSM) architectures invented by its founders. Its flagship product is Sonic, a family of ultra-low-latency text-to-speech and voice models exposed via API for use in voice agents, IVR, dubbing and assistants, with subsequent releases Sonic 2.0 and Sonic 3 improving naturalness, multilingual support and on-device performance.
Cartesia builds real-time multimodal AI models powered by state-space model (SSM) architectures invented by its founders. Its flagship product is Sonic, a family of ultra-low-latency text-to-speech and voice models exposed via API for use in voice agents, IVR, dubbing and assistants, with subsequent releases Sonic 2.0 and Sonic 3 improving naturalness, multilingual support and on-device performance.
Cartesia was founded in 2023 by Stanford researchers Karan Goel, Albert Gu, Arjun Desai, Brandon Yang, and their advisor Chris Ré, who together pioneered the Mamba state-space model architecture as a more efficient alternative to transformers. They started the company to commercialize SSMs for real-time AI, betting that lower latency and smaller compute footprint would unlock voice and other streaming workloads. Chris Ré's Stanford lab has also spawned SambaNova, Snorkel and Together AI.
Cartesia announced Sonic-3, its third-generation real-time voice model, alongside a $100M round backed by Kleiner Perkins, Index Ventures, Lightspeed and NVIDIA.
Cartesia released Mamba-3, an inference-first state-space-model architecture and accompanying research.
Company blog post announcing the Series A, Sonic 2.0 and the roadmap for real-time multimodal models.
Cartesia closed a $64M Series A to scale Sonic 2.0 and expand its real-time voice platform.
TechCrunch profile of Cartesia's state-space-model approach and its claim of running competitive voice models on-device.
Cartesia announced a $22M seed round to commercialize state-space-model architectures invented by its Stanford founding team.
Cartesia released Sonic, an SSM-based text-to-speech model targeting ultra-low-latency voice applications.
Co-founder & CEO
Previously PhD researcher at Stanford University
Stanford (PhD)
Stanford ML researcher; co-author of foundational state-space-model work. Co-founded Cartesia in 2023 and serves as CEO.
Co-founder & Chief Scientist
Previously Assistant Professor at Carnegie Mellon University
Stanford (PhD)
Co-inventor of the Mamba state-space model architecture and CMU assistant professor. Co-founded Cartesia in 2023.
Co-founder
Previously PhD researcher at Stanford University
Stanford (PhD)
Stanford ML researcher from the Chris Ré lab; co-founded Cartesia to commercialize state-space models.
Co-founder
Previously PhD researcher at Stanford University
Stanford
Stanford ML researcher; co-founded Cartesia in 2023 with labmates from the Chris Ré group.
Co-founder
Previously Professor at Stanford University
University of Washington (PhD)
Stanford CS professor and MacArthur Fellow; advisor and co-founder. His lab has also spawned SambaNova, Snorkel and Together AI.
$186M raised total