Tag
Volvo announced plans to use Nvidia's Hyperion platform and Alpamayo software to deliver hands-free, point-to-point Level 2+ driving across its lineup by 2030, combining Nvidia's AI stack with Volvo's safety and driver-monitoring expertise.
NVIDIA 宣布 RTX Spark 正式发布，并在全球多地举办 launch celebration 活动，主打面向创作者、玩家与 AI agent 的本地 AI PC 平台，多家媒体评测对其本地 AI 性能给予高度评价。
文章分析了AI数据中心液冷散热的投资机会：Jensen Huang在CES上透露NVIDIA下一代机柜将采用45°C热水冷却、取消冷凝器，液冷相关企业并购活跃（交易额超160亿美元），而两相冷却、防漏连接器和液冷运维仍是待开发的投资窗口。
OpenHands 宣布成为 NVIDIA Nemotron 3.5 Super VL 的发布合作伙伴，将这款开放多模态模型引入长期运行的软件工程智能体工作流，使智能体能够理解截图、仪表盘、文档和图表等非文本上下文。
A side-by-side benchmark comparison of Nemotron 3.5 Bolt, Nemotron 3.5 Super, and Qwen 3.8 (27B / FN) across tasks like SWE-bench Pro, Terminal Bench, GPQA-Diamond, and HLE, showing Qwen 3.8 FN leading on most metrics.
Nvidia's RTX Spark Arm-powered laptops launch October 16th, promising to run all Windows apps, games, and AI workloads via Microsoft's Prism emulator — aiming to avoid the fate of the failed Windows RT era.
NVIDIA's Nemotron 3.5 Super VL is now available on the River API for inference, finetuning, and RL, with claims that it outperforms Kimi K2.6 on text-to-SQL evaluations even before training and can be trained to beat Astra and Opus 5.5.
AfterQuery 宣布对 NVIDIA Nemotron 3.5 Super VL 进行金融信用任务后训练，与 Hebbia 合作将 Perfect solved 指标从 47% 提升至 74%，超过 Sol 6 并接近 Astra 6，成本约为其 1/7。
NVIDIA 推出开源权重的 Nemotron 3.5 Super VL 视觉语言模型（120B 总参数、12B 激活），通过 ChatNVIDIA 集成为 Deep Agents 带来视觉、推理与工具使用能力，并在行为套件与速度基准上表现突出。
NVIDIA 发布 Nemotron 3.5 Bolt 74B-A9B 24GB，这是一款采用原生 2-bit 量化、仅需 24GB 显存即可在本地运行日常智能体任务的视觉-语言模型，兼具速度与智能。
NVIDIA发布开放的Nemotron 3.5 Super VL（120亿参数级多模态模型）及两款边缘模型Bolt 74B-A9B 24GB与Bolt E4B 4GB，覆盖从数据中心到本地边缘设备的全场景智能体部署需求。
Spotify CEO Daniel Ek appeared on stage with Nvidia's Jensen Huang in Stockholm, discussing Europe's slow AI adoption and how AI is accelerating Spotify's internal engineering workflows.
NVIDIA released the Nemotron 3.5 Super VL 120B-A12B model family — a 120B-total/12B-active vision-language model with Mamba-2/MoE hybrid architecture, C-RADIO vision encoder, 1M-token context, and native speculative decoding via MTP — available in BF16, NVFP4, and other variants under the OpenMDW-1.1 license.
NVIDIA Vera Rubin NVL72 racks are beginning delivery, promising 4x efficiency, 2x speed, and doubled NVLink bandwidth, though questions remain about production adoption by Claude or GPT and how they compare to Google's custom chips.
NVIDIA 为 GeForce NOW 云游戏会员带来 DLSS 4.5 控制选项，让会员无需最新硬件即可自定义画质与性能平衡，同时《Enshrouded》1.0 与《Warhammer 40,000: Boltgun 2》等 10 款新游戏上线。
NVIDIA AI teases a busy week of AI model launches from various labs, hinting that it has one of its own models coming soon.
NVIDIA joins the European Commission's AI in Science Enablers Initiative at the AI in Science Summit in Dublin, pledging open models, datasets, compute, training resources, and collaborations to support Europe's scientific research community.
NVIDIA RTX 50 series GPU prices in Japan have surged dramatically, with the RTX 5060 Ti 16 GB rising up to 144% to $1,292 and the RTX 5080 climbing past $2,400, making cards nearly unaffordable in the region.
该推文回顾了OpenAI早期的一段轶事：英伟达发布首台AI超算DGX时无人问津，唯独埃隆·马斯克下单，但当时OpenAI是一家非营利机构。
用户提到6月购买的GPU和DGX Spark价格至少上涨了1.5倍，并指出本地AI加速的性能数据很明确。