Nvidia launches Nemotron 3 Super with 5× throughput, seals Palantir AI pact
Nvidia launched its Nemotron 3 Super AI model featuring 120 billion parameters and a one-million-token context window to support complex agentic tasks. The model delivers up to 5× higher throughput and twice the accuracy versus Nemotron Super and is being integrated by Palantir, Cadence, Dell and Hewlett Packard Enterprise.
1. Nemotron 3 Super Launch
Nvidia introduced Nemotron 3 Super, its latest open AI model designed for agentic systems that can autonomously reason, plan and execute multi-step workflows. The rollout underscores Nvidia’s push beyond simple language tasks toward fully autonomous AI agents.
2. Enhanced Performance
The model leverages a hybrid mixture-of-experts architecture with 120 billion parameters—only 12 billion active during inference—and a one-million-token context window. This design yields up to 5× higher throughput and twice the accuracy compared with the previous Nemotron Super.
3. Strategic Enterprise Partnerships
Several enterprise and technology firms, including Palantir, Cadence, Dell Technologies and Hewlett Packard Enterprise, are integrating Nemotron 3 Super into cybersecurity automation, semiconductor design and AI workflow platforms, aiming to customize the system for their specialized workloads.