Model Release
Nvidia's Nemotron 3 Ultra Becomes Smartest Open US Model
Nvidia's Nemotron 3 Ultra is the most capable open AI model from the US, scoring 48 on Artificial Analysis intelligence ranking as of June 1, 2026.
Photo: Tara Winstead / Pexels
Nvidia has released its Nemotron 3 Ultra, which is now the most capable open AI model from the United States, according to benchmark platform Artificial Analysis. The model has a total of 550 billion parameters, with approximately 55 billion active at any given time. On the Artificial Analysis intelligence ranking, Nemotron 3 Ultra scores 48 points, significantly outperforming other open US models such as Gemma 4 31B (39), Nemotron 3 Super (36), and gpt-oss-120b (33). However, it does not reach the top open models from China, such as Kimi K2.6, which scores 54 points. The current strongest closed model, Opus 4.8, scores 61 points. According to Artificial Analysis, Nemotron 3 Ultra lands in the 'most attractive quadrant' of the chart, combining high intelligence scores with fast output speed. On provider DeepInfra, Nemotron 3 Ultra delivers more than 300 tokens per second, compared to 50 to 100 for similarly sized models from DeepSeek or Moonshot. Nvidia announced the model will be released on June 4 on platforms such as Hugging Face and OpenRouter. *Source: [thedecoder](https://the-decoder.com/nvidias-nemotron-3-ultra-becomes-the-smartest-open-us-model-but-china-still-leads/)*
Key points
- Nvidia's Nemotron 3 Ultra is the most capable open AI model from the US, according to Artificial Analysis.
- Nemotron 3 Ultra has a total of 550 billion parameters, with about 55 billion active at any given time.
- Nemotron 3 Ultra scores 48 points on the Artificial Analysis intelligence ranking, surpassing Gemma 4 31B (39), Nemotron 3 Super (36), and gpt-oss-120b (33).
- Kimi K2.6 scores 54 points on the Artificial Analysis intelligence ranking, outperforming Nemotron 3 Ultra.
- Opus 4.8, the current strongest closed model, scores 61 points on the Artificial Analysis intelligence ranking.
- Nemotron 3 Ultra lands in the 'most attractive quadrant' of the Artificial Analysis chart, combining high intelligence scores with fast output speed.
- Nemotron 3 Ultra delivers more than 300 tokens per second on DeepInfra, compared to 50 to 100 for similarly sized models from DeepSeek or Moonshot.