STORY · MODELLER_
NVIDIA strikes back with open 120B model – Nemotron 3 Super sets new standard for agentic AI
NVIDIA launched Nemotron 3 Super, a 120B-parameter model with hybrid Mamba-Transformer architecture and 1M context window that is 2.2x faster than competitors. The model is unusually open with released code, training data and infrastructure details, and received immediate support from major frameworks like vLLM and Ollama.
WHY IT MATTERS
This marks an important shift in the AI industry from isolated models toward agent systems and orchestration. The combination of NVIDIA's technical innovations (MTP, KV-cache efficiency) and growing focus on agent runtimes signals that competition is now about infrastructure and integration as much as model quality.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.