CAMB.AI launches MARS8, the first family of text-to-speech models built for real-world deployment, enabling enterprises to run the right voice architecture for speed, quality, or efficiency at scale on their own infrastructure
SAN FRANCISCO, CA / ACCESS Newswire / January 20, 2026 / CAMB.AI, the AI voice infrastructure company founded by AI researchers from Apple and Carnegie Mellon, today announced the launch of MARS8. This is the first text-to-speech (TTS) system designed not as a single model, but as a family of specialized architectures built for production reality.
Since 2023, CAMB.AI’s MARS engine has powered AI voice-enabled applications reaching over 200 million users through partners like NASCAR, Broadcom, Australian Open, Eurovision Sport, IMAX, and Comcast NBCUniversal.
Powering Enterprise Scale Since 2023
Past viral launch and success of MARS5 in the open-source domain
After shipping voice AI into high-stakes production environments across 150+ languages, the team uncovered a critical flaw in the current market: No single TTS architecture wins across every use case.
“The market forces developers to choose between speed, quality, accuracy, and cost. We realized that was a false choice,” says Akshat Prakash, CTO at CAMB.AI. “A live voice assistant needs sub-150ms latency. A movie dubbing pipeline needs director-level emotional control. An automotive system has strict memory constraints. You cannot solve all three with one generic API.”
Introducing the MARS8 Family
MARS8 moves away from the “black box” API model by offering four distinct architectures, each optimized for specific production constraints:
MARS-Flash: Ultra-low latency (sub-150ms TTFB) designed for real-time agents and call centers. Optimized for Blackwell GPUs, L4 and L40S
MARS-Pro: The workhorse for expressive dubbing and digital media, balancing fidelity with speed.
MARS-Instruct: Director-level control for high-end film production, allowing independent tuning of speaker and prosody.
MARS-Nano: A highly efficient 50M parameter model for on-device applications where compute is constrained.
Note: More information on the MARS8 family towards the end of the document.
A Compute-First Business Model
In a move set to disrupt the unit economics of Voice AI, MARS8 is launching with a “Compute-First” model. The status quo of “pay-per-character” API pricing destroys margins at scale. CAMB.AI is flipping the script.
MARS8 allows enterprise customers to run the models on their own infrastructure – whether that is AWS Bedrock, Google Cloud Vertex AI, or specialized GPU platforms like Modal, Baseten and others. MARS8 is launching on 25+ compute platforms and on-device SDKs which marks a historic first in the Voice AI space.
Example of MARS8 ready for deployment on Google Cloud Model Garden, similar to Anthropic
“Token-based pricing scales linearly. Compute-based pricing flatlines costs even as usage spikes, saving enterprises up to 90%,” the company states. “Plus, by deploying in your own region, you control the latency floor and keep data within your compliance rails.”
Launch Ecosystem
MARS8 is launching with a robust partner ecosystem, bridging the gap between infrastructure and application:
Compute Partners: Ensuring MARS8 runs on the hardware developers already use (Google Cloud, AWS, Azure (coming soon), Hathora, Modal, Baseten and others).
Voice Agent Platforms: Integrating MARS-Flash directly into the workflow of top conversational AI providers for immediate deployment in contact centers.
MARS8 marks a turning point for production Voice AI. The days of forcing every use case through a single generic model are over. Whether you’re shipping a real-time voice agent, dubbing a feature film, or building an on-device assistant, you now have architecture built for your constraints, not compromises. And with compute-first pricing, scaling no longer means bleeding margin.
Ready to deploy? Explore the full MARS8 family at camb.ai/mars. For technical specs and architecture deep dives, check out camb.ai/blog-post/mars8-technical-report.
About CAMB.AI
CAMB.AI is the localization AI infrastructure company built to help the world’s biggest enterprises reach 8 billion consumers. Backed by real-world usage in live sports, cinema, and broadcasting, CAMB.AI is pioneering real-time, language-free access across text, audio, and video.
SOURCE: CAMB.AI
