Nemotron 3 is not just a model, it is a complete agentic AI development stack. Torque is the governed infrastructure layer that makes every part of it enterprise-ready, from NIM inference to NeMo fine-tuning to reinforcement learning at scale.
Today, May 5th 2026, Quali announced Torque’s support of the NVIDIA Nemotron 3 family of open models, extending Torque’s governed infrastructure capabilities to cover the full NeMo and NIM agentic AI stack. The announcement coincides with the general availability of Nemotron 3 Super and the upcoming release of Nemotron 3 Ultra, giving enterprises a production-ready path to deploy, manage, and scale Nemotron 3 workloads from day one.
Nemotron 3 ships with NIM microservices for production inference, NeMo fine-tuning pipelines, and reinforcement learning environments purpose-built for multi-agent systems at scale. Each layer of that stack demands governed GPU infrastructure to deliver real enterprise value. Torque provides exactly that, zero-touch lifecycle management, self-service access for data scientists and ML engineers, and consistent governance across the full stack, from desk-side DGX Spark hardware to on-premises clusters and cloud GPU infrastructure.
BULLET POINTS:
- Governed NIM Deployment: Torque provisions and lifecycle-manages NVIDIA NIM microservice environments via automated GPU Operator and NIM Operator integration, enabling teams to deploy Nemotron 3 inference services from versioned blueprints with full policy enforcement and automatic teardown.
- Self-Service for AI Teams: Data scientists and ML engineers access governed Nemotron 3 fine-tuning and NeMo RL training environments on demand through Torque’s self-service portal, no infrastructure expertise required, no IT tickets, with cost attribution and compliance controls built in.
- Unified Control Plane Across Hardware: The same Torque blueprints and governance policies that manage Nemotron 3 workloads on NVIDIA DGX Spark and DGX Station scale seamlessly to on-premises GPU clusters and cloud infrastructure, giving enterprises a consistent operational model across their entire AI estate.
- Broadest NVIDIA Ecosystem Coverage: Torque’s support for Nemotron 3 extends validated coverage across DGX Spark, DGX Station, NVIDIA GPU Operator, NIM Operator, NemoClaw, and now the complete Nemotron 3 model family from Nano through Ultra, with Quali among the earliest software partners to validate Torque on NVIDIA’s Blackwell architecture.
