NVIDIA Unveils Open-Source Nemotron 3 Super Model, Rivaling Top AI Performance

NVIDIA has once again made a significant impact in the AI large model arena. On March 12, the company officially unveiled a new generation of open-source large language model specifically designed for AI agents: Nemotron 3 Super . With its exceptional reasoning efficiency and impressive task success rate, this model quickly became a focal point within the open-source community.
Architectural Innovation: 300% Faster Inference Speed
Nemotron 3 Super employs an innovative Mamba-MoE hybrid architecture, featuring 120 billion total parameters with only 12 billion activated per inference. This design enables it to maintain robust performance while tripling inference speed and quintupling throughput. Furthermore, the model supports an ultra-long context window of up to 1 million tokens, effectively addressing common challenges in multi-agent collaboration like "goal drift" and "context explosion."
Performance Benchmark: The New "Performance Ceiling" for Open-Source Models
In multiple authoritative benchmarks, Nemotron 3 Super delivered outstanding results. It not only topped the efficiency and openness rankings from Artificial Analysis but also propelled NVIDIA's self-developed AI-Q agent to first place on both DeepResearch Bench leaderboards. Notably, the model achieved an 85.6% success rate on the popular agent task OpenClaw, with performance approaching that of leading closed-source models like Claude Opus 4.6 and GPT-5.4.
Blackwell Platform Compatibility: Native Support for NVFP4 Training
To fully leverage its proprietary hardware advantages, Nemotron 3 Super supports not only BF16 and FP8 formats but also introduces specific support for NVFP4 training on NVIDIA's latest Blackwell platform and subsequent architectures. This capability is poised to further reduce large model training costs and improve computational efficiency.
Ecosystem Adoption: Major Industry Players Are On Board
Currently, Nemotron 3 Super has already been integrated by several technology leaders, including Perplexity, Palantir, Siemens, and Dell, and is available on major cloud platforms like AWS, Azure, and Google Cloud. As a free, open-source model, it provides developers with a powerful, cost-effective alternative, significantly challenging the current market dominance of closed-source large models.
Related article
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage
California AV Compliance: A New Era of Tickets, Geofences, and 1M Miles
Guident operates an AuveTech shuttle in South Florida, managing a four-mile route in West Palm Beach and a one-mile route in Boca Raton using its remote monitoring technology. | Credit: GuidentCalifornia is redefining the regulatory landscape for dri
Related Special Topic Recommendations
Comments (0)
0/500

NVIDIA has once again made a significant impact in the AI large model arena. On March 12, the company officially unveiled a new generation of open-source large language model specifically designed for AI agents:
Architectural Innovation: 300% Faster Inference Speed
Performance Benchmark: The New "Performance Ceiling" for Open-Source Models
In multiple authoritative benchmarks,
Blackwell Platform Compatibility: Native Support for NVFP4 Training
To fully leverage its proprietary hardware advantages,
Ecosystem Adoption: Major Industry Players Are On Board
Currently,
DeepMind CEO Hassabis: I sleep six hours a day, usually feel energetic around 1 a.m.
Fortune recently featured an interview with Demis Hassabis, CEO of Google DeepMind, revealing his unconventional approach to rest and productivity. Hassabis disclosed that he sleeps very little, structuring his waking hours into two distinct work blo
OpenAI, Anthropic Vie for Market Share Despite Revenue Shortfalls
Despite recent reports suggesting OpenAI missed revenue targets, creating pressure on tech stocks this Tuesday, private AI lab investors remain resilient. Seasoned backers have confirmed they will not reduce investment despite negative media coverage





Home






