Devops

MLCommons Releases MLPerf Inference v5.1, Setting New AI Performance Benchmarks MLCommons has released MLPerf Inference v5.1, the latest version of its benchmark suite, introducing new tests to evaluate AI system performance on modern workloads. This update saw a record 27 organizations submitting systems for evaluation. The new version features three new benchmarks, including one for the DeepSeek-R1 reasoning model, and implements tighter latency requirements for certain large language model tests. Notably, the popular Llama 2 70B benchmark demonstrated significant performance gains, with top systems running up to 50% faster than in the previous release. The v5.1 release also marked the first submission of a heterogeneous system that intelligently balances inference workloads across different accelerator types. ...

OpenAI Taps Thinking Machines as First APAC Services Partner to Boost Enterprise AI Thinking Machines Data Science is now OpenAI’s first official services partner in the Asia-Pacific (APAC) region. This strategic collaboration aims to accelerate enterprise AI adoption across APAC. Thinking Machines will provide businesses with specialized training for ChatGPT Enterprise, support for building custom AI applications, and expert guidance on integrating AI into daily operations. The partnership will initially target Singapore, the Philippines, and Thailand, with future plans to expand into other markets and develop tailored services for key sectors like finance, retail, and manufacturing. This initiative directly addresses the challenge many regional companies face in converting AI adoption into tangible business value. ...

Tech News Roundup: Anthropic, Quantinuum, and Kubernetes Updates This week’s technology digest highlights major developments across the industry, from a landmark legal settlement in AI that could reshape how models are trained, to massive funding in quantum computing, and crucial updates for cloud-native infrastructure and open-source software. Anthropic’s $1.5B Settlement Over AI Training Data Sets Industry Precedent AI company Anthropic has agreed to a $1.5 billion settlement in a major lawsuit concerning the use of pirated books to train its AI models. The lawsuit, filed by authors, alleged that the company used their copyrighted works without permission to develop its chatbots. This settlement represents a significant moment for the AI industry and could establish new standards regarding the use of copyrighted materials for training large language models (LLMs). ...

Gartner’s 2025 Hype Cycle: Generative AI Enters ‘Trough of Disillusionment’ According to Gartner’s 2025 Hype Cycle for Artificial Intelligence, Generative AI has officially moved into the ‘Trough of Disillusionment’. This shift indicates that the initial peak of inflated expectations is cooling as organizations now grapple with the technology’s limitations and seek tangible returns on investment (ROI). Gartner notes that while GenAI remains a potentially transformational technology, the primary challenge for many companies is proving its business value. Consequently, the industry’s focus is shifting towards foundational elements like AI-ready data and the development of AI agents, both of which are currently at the ‘Peak of Inflated Expectations’. This trend underscores a maturing understanding that successful enterprise AI relies on strong data governance, intelligent workflows, and robust operational frameworks. Despite cooling expectations for GenAI, overall IT spending is still projected to increase in 2025, driven partly by ongoing investments in AI software and infrastructure. ...

Daily News Feed: AI Security, Quantum Leaps, and Microsoft’s Rust Bet Cloudsmith Launches ML Model Registry for Centralized AI Model Management Cloudsmith, a software supply chain management company, has launched its ML Model Registry. This new platform will serve as a single source of truth for all AI models and datasets that a company utilizes. The registry integrates with the Hugging Face Hub and SDK, allowing developers to efficiently push, pull, and manage their machine learning models. ...

Mount Sinai’s ‘AEquity’ AI Tool Aims to Eliminate Healthcare Algorithm Bias Researchers at the Icahn School of Medicine at Mount Sinai have introduced AEquity, a novel AI tool designed to detect and mitigate biases in healthcare machine learning algorithms. As detailed in the Journal of Medical Internet Research, this tool confronts the critical challenge of algorithmic bias, which can compromise the accuracy of medical diagnoses and treatment strategies. AEquity was validated on diverse health datasets, including medical imagery and patient records, successfully identifying both known and previously undiscovered biases. The primary objective of AEquity is to ensure AI-driven healthcare solutions are both effective and equitable for all patient demographics, particularly those underrepresented in training data. The framework is built for adaptability, supporting a wide array of machine learning models. ...

O2 Telefonica, Tech Mahindra & Nvidia to Build Generative AI for Telecom Operations O2 Telefonica is partnering with Tech Mahindra and Nvidia to develop a generative AI model designed to revolutionize telecom network operations. The “Large Telco Model” (LTM) aims to significantly improve operational efficiency and customer service reliability. This advanced system will empower technicians to identify network disruptions faster and coordinate field operations more effectively. Built on Nvidia’s AI Enterprise software, the model leverages agent-based AI specifically tailored for the telecommunications sector. This initiative is a key part of O2 Telefonica’s strategy to pioneer data-driven and autonomous network management in Germany. ...

California Colleges Partner with Google & Microsoft for Free AI Training California’s public university and community college systems are launching a major initiative with tech giants like Google and Microsoft to offer free artificial intelligence training to students. This partnership aims to address the evolving job market, equipping the future workforce with critical AI skills as automation impacts entry-level roles. Governor Gavin Newsom emphasized that the state’s public higher education system is the backbone of economic development. In exchange for providing free AI resources, the tech companies gain a direct channel to millions of potential new users. The program will also include AI training for instructors across California’s 116 community colleges, which serve over 2 million students, meeting the rising employer demand for AI proficiency in tech and non-tech sectors alike. ...

OpenAI Unveils GPT-Realtime: A Leap Forward in Speech-to-Speech AI OpenAI has officially launched GPT-Realtime, its most advanced speech-to-speech model, and has moved its Realtime API out of beta with new enterprise-grade features. GPT-Realtime is engineered to understand and generate audio directly, which significantly reduces latency and preserves crucial nuances like emotion and laughter often lost in traditional text-to-speech systems. The new model showcases enhanced performance in reasoning, instruction following, and function calling. The updated Realtime API now supports image input, Session Initiation Protocol (SIP) for phone network connectivity, and Multi-Capability Pipelines (MCP) for pluggable features. OpenAI also introduced two new voice options, Cedar and Marin, and has cut the pricing for the Realtime API by 20%. ...

Microsoft Challenges OpenAI with In-House MAI-1 and MAI-Voice-1 AI Models Microsoft has begun public testing for MAI-1, its first internally developed, large-scale AI model, marking a significant strategic move to reduce its reliance on partner OpenAI. The new foundational model is engineered for general-purpose tasks, including answering complex questions and following instructions. Alongside MAI-1, Microsoft also unveiled MAI-Voice-1, a sophisticated model capable of generating natural-sounding, human-like speech. The voice model can reportedly generate one minute of audio in under a second using minimal computing power. These releases signal escalating competition in the generative AI space, especially within multimodal and voice-based applications. ...