NVIDIA Sets Llama Speed Records Alongside Amazon Launches Kindle Translate

The latest MLPerf Training v5.1 results, released on November 12, 2025, reveal rapid advancements in AI, particularly for generative AI tasks, with 65 unique systems from 20 organizations participating. NVIDIA stands out as a dominant force, winning all seven tests and setting new speed records. Its new GB300 NVL72 system, powered by Blackwell Ultra GPUs, demonstrated remarkable performance gains, proving 4 times faster for Llama 3.1 405B pretraining and 5 times faster for Llama 2 70B LoRA fine-tuning compared to the previous Hopper architecture. NVIDIA also pioneered the use of FP4 precision for calculations, significantly boosting performance while maintaining accuracy, even training Llama 3.1 405B in just 10 minutes using over 5,000 Blackwell GPUs. Complementing this, Lambda's NVIDIA GB300 NVL72 system showed a 27 percent speedup over the previous GB200 NVL72 for Llama 2-70B LoRA and a 13 percent speedup using FP4 precision for Llama 3.1 8B.Beyond benchmarks, AI's influence spans various sectors. Cisco is actively preparing partners for the AI era by focusing on security, a critical need given that 86 percent of organizations faced an AI-related security problem last year. The company launched Cisco Unified Edge on November 3, a platform combining security, computing, networking, and storage for deploying AI at the network's edge, and updated its 360 Partner Program with new specializations. In healthcare, AI is transforming patient care nationwide, assisting doctors in spotting serious conditions like sepsis, conducting neurological research, and improving patient safety by reducing misdiagnoses and preventable medical errors.The push for AI literacy and skill development is also gaining momentum. AWS launched new training and certification programs on November 12, including Meeting Simulator and Cohorts Studio, and will open registration for a new AWS Certified Generative AI Developer Professional certification beta exam on November 18. Universities, such as the University of North Carolina at Chapel Hill, are teaching students how to use generative AI tools ethically and effectively, while restaurants explore AI to enhance employee skills.Meanwhile, Amazon is leveraging AI for its Kindle Translate beta program, allowing Kindle Direct Publishing authors to translate eBooks into multiple languages within 72 hours, initially supporting English-Spanish and German-English. In e-commerce, ShipToBox.com utilizes AI for its Smart Global Shopping & Shipping Platform, helping customers worldwide save up to 10 percent on US sales tax and optimize shipping for Black Friday purchases from major US stores like Amazon. However, the widespread use of AI for content creation raises concerns, with some arguing that it leads to generic, unoriginal blog content, despite its utility for tasks like summaries and meta descriptions.

Key Takeaways

• MLPerf Training v5.1 results, released November 12, 2025, highlight significant advancements in generative AI, with 65 unique systems from 20 organizations participating.
• NVIDIA dominated the MLPerf benchmarks, winning all seven tests with its GB300 NVL72 system, achieving 4 times faster Llama 3.1 405B pretraining and 5 times faster Llama 2 70B LoRA fine-tuning compared to its previous Hopper architecture.
• NVIDIA was the first to use FP4 precision in MLPerf, setting new records, including training Llama 3.1 405B in just 10 minutes using over 5,000 Blackwell GPUs.
• Lambda's NVIDIA GB300 NVL72 system showed a 27 percent speedup for Llama 2-70B LoRA compared to the previous GB200 NVL72 and a 13 percent speedup using FP4 precision for Llama 3.1 8B.
• Cisco launched Cisco Unified Edge on November 3, a secure AI edge platform, and updated its partner programs to address AI security, as 86 percent of organizations faced an AI-related security problem last year.
• AI is transforming healthcare by aiding in the detection of serious conditions like sepsis, supporting neurological research, and improving patient safety by reducing misdiagnoses and medical errors.
• AWS is addressing the AI skills gap with new training programs like Meeting Simulator and Cohorts Studio, and will open registration for a new AWS Certified Generative AI Developer Professional certification beta exam on November 18.
• Amazon introduced Kindle Translate, an AI-powered beta service for Kindle Direct Publishing authors to translate eBooks into multiple languages within 72 hours, initially between English and Spanish, and from German to English.
• Universities are teaching AI literacy and ethical use, while the content industry debates AI's impact on blog quality, noting its utility for summaries but potential for generic content.
• ShipToBox.com uses AI for global Black Friday shopping, enabling customers in over 220 countries to save up to 10 percent on US sales tax and optimize shipping from US stores like Amazon.

MLCommons reveals new AI training benchmark results

MLCommons released the MLPerf Training v5.1 results on November 12, 2025, showing rapid improvements in AI, especially for generative AI tasks. This benchmark round featured 65 unique systems from 20 organizations, including new participants like Datacrunch and the University of Florida. Performance for generative AI benchmarks such as Llama 2 70B LoRA and Llama 3.1 8B significantly improved, outpacing Moore's Law predictions. The new version also updated its tests, replacing older ones with Llama 3.1 8B and Flux.1 to better reflect current AI technology.

NVIDIA dominates MLPerf AI training benchmarks

NVIDIA won all seven tests in the MLPerf Training v5.1 benchmarks, setting the fastest training times for various AI tasks. The new GB300 NVL72 system, powered by NVIDIA Blackwell Ultra GPUs, showed huge performance gains. It was 4 times faster for Llama 3.1 405B pretraining and 5 times faster for Llama 2 70B LoRA fine-tuning compared to the previous Hopper architecture. NVIDIA also became the first to use FP4 precision for calculations, which significantly boosted performance while maintaining accuracy. The company set new records, including training Llama 3.1 405B in just 10 minutes using over 5,000 Blackwell GPUs.

Lambda's new NVIDIA system speeds up AI training

Lambda announced its MLPerf Training v5.1 results, showing its NVIDIA GB300 NVL72 system is very powerful for training large language models. Lambda's system was 27 percent faster than the previous GB200 NVL72 for Llama 2-70B LoRA. It completed the Llama 2-70B LoRA test in 1.26 minutes, which is 1.27 times faster than Oracle's best GB200 NVL72 result from the last round. These gains come from the NVIDIA Blackwell Ultra architecture, updated software, and Lambda's special cluster design. Lambda also noted a 13 percent speedup using FP4 precision compared to FP8 for Llama 3.1 8B.

Cisco helps partners secure AI growth

Cisco is helping its partners prepare for the AI era by focusing on security. A recent Cisco study shows only 14 percent of organizations feel ready for AI, and 86 percent faced an AI-related security problem last year. Cisco announced new tools and programs at Cisco Partner Summit and Cisco Live Melbourne to help. These include ways to modernize networks with built-in security using Cisco Access Manager. Cisco also updated its 360 Partner Program with new specializations and incentives to boost partner profits in AI and security.

Cisco offers partners new secure AI edge platform

Cisco is giving its partners a new way to use AI at the edge of networks, where data is created. The company launched Cisco Unified Edge on November 3, which is now available to order. This platform combines security, computing, networking, and storage into one easy-to-use system. It helps partners quickly deploy AI in many locations like retail stores and hospitals. Cisco Unified Edge offers simple setup, built-in security, and central management through Cisco Intersight. Cisco is also supporting partners with training webinars, technical roadshows, and co-marketing efforts to help them succeed with this new AI opportunity.

AI changes healthcare nationwide

Artificial Intelligence is changing healthcare across the country, helping doctors in many ways. Hospitals are using AI more and more to spot serious conditions like sepsis and to conduct neurological research. This technology is expanding its uses in healthcare settings. Dr. Akshay Sayal from NBC News reported on these changes from Cleveland, Ohio, on November 13, 2025.

AWS launches new AI training and certification

AWS is helping people build AI careers by offering new training and certification programs. On November 12, AWS Skill Builder launched Meeting Simulator, Cohorts Studio, and microcredentials to help close the AI skills gap. These tools offer hands-on learning, including AI-powered communication practice and group study. A new AWS Certified Generative AI Developer Professional certification beta exam will open for registration on November 18. AWS Skill Builder also provides over 220 free AI courses, from beginner to advanced topics.

Amazon Kindle Translate offers fast AI eBook translations

Amazon is launching Kindle Translate, a new service that uses AI to translate eBooks for Kindle Direct Publishing authors. This beta program aims to help authors share their books globally in multiple languages within 72 hours. Initially, it supports translations between English and Spanish, and from German to English. Amazon says the AI translations are automatically checked for accuracy with help from language experts. Readers will know if a book is AI-translated and can preview samples.

University experts teach students AI literacy

Dayna Durbin, a librarian, and Dan Anderson, an English professor at the University of North Carolina at Chapel Hill, are teaching students about AI literacy. They help students understand how to use generative AI tools ethically and effectively for their studies and future careers. While AI writing can be "flat," it is useful for brainstorming ideas and summarizing research. A workshop on AI and misinformation led by Durbin attracted 90 students, showing a strong interest in the topic. They believe it is important to provide resources as students will use these tools in their jobs.

AI can improve patient safety in healthcare

Dr. Robert Pearl argues that using generative AI in healthcare will help patients, not harm them. He points out that many people die each year from misdiagnoses and preventable medical errors in the US. AI can provide timely guidance to patients and alert doctors to early warning signs, helping to adjust medications before problems worsen. It can also offer support for mental health conditions when clinicians are not available. Dr. Pearl believes combining dedicated clinicians, empowered patients, and AI will lead to safer, better, and more accessible healthcare for everyone.

AI writing hurts blogs and content quality

The author believes that using AI for full blog writing is a bad idea, even though it is helpful for tasks like summaries, meta descriptions, and alt text. AI models often use similar training data, leading to generic and unoriginal blog content that does not add unique value for readers. This trend also hurts the blogging industry, as people increasingly go directly to chatbots for information instead of reading blogs. While AI content is cheap to produce, it contributes to a cycle of low-quality information online.

ShipToBox uses AI for global Black Friday shopping

ShipToBox.com is using AI to make Black Friday shopping easier and more affordable for customers worldwide. Their Smart Global Shopping & Shipping Platform, powered by PayToMe.co, allows shoppers in over 220 countries to buy from top US stores like Amazon and Nike with one checkout. Customers can save up to 10 percent on US sales tax and compare courier prices for global delivery. The AI SmartCart automatically finds deals and suggests the best shipping options. ShipToBox.com also won the 2025 Stevie International Business Award for its innovative use of AI in logistics.

Restaurants can use AI to improve employee skills

Restaurants can use AI to improve efficiency and help employees learn new skills. Yutaka Takagi from isolved explains that traditional training programs are often outdated and do not adapt to new tech trends like AI. Hospitality leaders should create tailored training programs that teach AI literacy and ethics, recognizing different learning styles. These programs should integrate into daily workflows, allowing teams to experiment and learn safely. By doing this, restaurants can close the AI skills gap and help employees thrive alongside new technology.

Sources

• MLCommons Releases MLPerf Training v5.1 Results
• NVIDIA Wins Every MLPerf Training v5.1 Benchmark
• MLPerf Training v5.1: Lambda’s NVIDIA GB300 NVL72 outperforms GB200 by 27%
• Connected, Protected, Ready: Securing Partner Growth in the AI Era
• Beyond the Cloud: Partners’ Secure AI Edge Opportunity with Cisco
• How AI is transforming healthcare across the country
• Build your AI career path with AWS's new certification and hands-on learning experiences
• Kindle Translate: How Amazon Wants to Create AI-Translated eBooks in 72 Hours
• Librarian and English professor build AI literacy
• Rejecting generative AI in healthcare won’t protect patients – it will harm them
• Writing is probably the worst use case of AI
• ShipToBox.com Powers Black Friday AI Shopping -- One Checkout, Save on U.S. Sales Tax
• Can Restaurants Leverage Effective AI Upskilling Programs?