[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-63255594-b16a-4106-9acc-2dc479b97e14":3},{"id":4,"title":5,"summary":6,"original_url":7,"source_id":8,"tags":9,"published_at":23,"created_at":24,"modified_at":25,"is_published":26,"publish_type":27,"image_url":13,"view_count":28},"63255594-b16a-4106-9acc-2dc479b97e14","英伟达Nemotron 3 Ultra登场：550B开源模型刷新美国开放权重智能榜单","在6月1日台北Computex 2026主题演讲中，英伟达CEO黄仁勋正式发布了Nemotron 3 Ultra——一款拥有约5500亿参数的开源大模型，一举成为美国开放权重模型中的智能新标杆。\n\nNemotron 3 Ultra采用90%稀疏性的MoE（专家混合）架构，实际激活参数约550亿。在Artificial Analysis的智能指数评测中，该模型取得48分，大幅领先Gemma 4 31B（39分）、Nemotron 3 Super（36分）以及gpt-oss-120b（33分），不过仍略低于中国主导的开源前沿模型Kimi K2.6（54分）。\n\n性能方面，DeepInfra内测端点显示Nemotron 3 Ultra推理速度超过300 tokens\u002F秒，而同量级的中国模型（DeepSeek、Moonshot等）通常在50-100 tokens\u002F秒区间。速度优势来自NVFP4量化支持，这与此前Nemotron 3 Super的思路一脉相承。\n\n值得关注的是，Nemotron 3 Ultra定位为「Agentic Coding & Search」场景，这意味着英伟达正将开源大模型从通用对话推向企业级自动化工作流。随着美国开放权重模型在智能和速度上同时逼近闭源前沿，2026年开源与闭源的差距将进一步收窄。","https:\u002F\u002Fblogs.nvidia.com\u002Fblog\u002Fnemotron-3-ultra\u002F","474eef8c-e0c3-46cf-adee-c089558220f9",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"0a93ec8e-ea39-4693-81de-563ca8c173f7","inference",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":18,"name":19,"slug":19,"description":13,"color":13},"8dac812d-3839-4abe-a855-5f56ec9515fd","nvidia",{"id":21,"name":22,"slug":22,"description":13,"color":13},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source","2026-06-01T10:10:00Z","2026-06-01T10:05:43.627369Z","2026-06-01T10:05:43.627379Z",true,"agent",3]