Nvidia brings Nemotron models to ElevenLabs voice agents, starting with Poland
In brief
- Nvidia's Nemotron models are coming to ElevenAgents, ElevenLabs' voice-agent platform for enterprise and government customers.
- ElevenLabs' New York Summit was the venue for the November 11, 2026, announcement.
- Poland's e-Health platform expects about 40 million voice-agent bookings a year, the companies said.
- Nemotron 3.5 Super VL cuts costs 70% on key benchmarks, according to the companies' own figures.
- ElevenLabs also added agent memory, routines, a partner program and customization options.
What's in the deal
The deal centers on Nemotron 3.5 Super VL, a family of open Nvidia models built for reasoning in agentic workloads. ElevenLabs contributes the voice side (its speech technology is described as low-latency and multilingual), while Nvidia's models handle the reasoning inside ElevenAgents.
Nvidia brings the reasoning. ElevenLabs brings the voice.
The companies' announcement materials claim Nemotron 3.5 Super VL achieves 70% lower costs on key benchmarks compared with leading models. That's the companies' own figure, not an independent benchmark result, and the report doesn't name the research it's drawn from.
The partnership is also set to add 15 MW of compute capacity through Nvidia's Blackwell Ultra systems and the Vera Rubin platform, according to the announcement as reported by Crypto Briefing.
Poland's e-Health platform goes first
So what does this look like in practice? A voice agent is being built for Centralna e-Rejestracja, Poland's national e-Health platform, to help patients schedule medical appointments. Patients will be able to book through landlines, and the platform is set to handle approximately 40 million appointment bookings annually using voice agents, per the companies' announcement.
Neither CEO was quoted directly in the report (Crypto Briefing paraphrased both). ElevenLabs CEO Mati Staniszewski framed the expansion around security, saying secure operational frameworks become necessary as voice agents function as essential infrastructure. Nvidia CEO Jensen Huang took a wider view, pointing to the potential of computers understanding human language at a national scale.
New ElevenAgents features
ElevenLabs also introduced agent memory, routine capabilities, a partner program and customization options driven by customer conversation data. The memory feature lets an agent retain context across interactions, so it isn't starting from zero on every call.
ElevenLabs was founded in 2022. Crypto Briefing reports it's valued at approximately $22 billion following a strategic investment from Nvidia in 2025, which means Nvidia was already a backer before the Nemotron expansion.
Frequently asked questions
What is Nemotron 3.5 Super VL?
Nemotron 3.5 Super VL is a family of open Nvidia models built for reasoning in agentic workloads. It's the core of the expanded deal with ElevenLabs. The companies' announcement materials claim it achieves 70% lower costs on key benchmarks compared with leading models, but that isn't an independent benchmark result.
How will Poland use the Nvidia and ElevenLabs voice agents?
A voice agent is being built for Centralna e-Rejestracja, Poland's national e-Health platform, to help patients schedule medical appointments. Patients will be able to book through landlines. According to the companies' announcement, the platform is set to handle approximately 40 million bookings annually using voice agents.
What new features did ElevenLabs add to ElevenAgents?
ElevenLabs introduced agent memory, routine capabilities, a partner program and customization options driven by customer conversation data. The memory feature lets an agent retain context across interactions.
