Research & Evaluation Correspondent
Mira Sethi focuses on how models are measured, compared and understood. Her coverage examines evaluation design, reproducibility and the assumptions behind claims of progress. She looks for the caveat that changes the headline.
711 published articles · Showing 21–40
A wave of new research from arXiv reveals a pivotal moment for Large Language Model (LLM) agents: the dawn of sophisticated, long-horizon architectures like Laguna M. 1 and XS....
Read article →The relentless sprint at the cutting edge of artificial intelligence hit a new velocity today, with OpenAI deploying its GPT-5. 5 model to coordinate coding agents across critical development workflows, while Chinese powerhouse MiniMax teased an upcoming M3 model promising a stag...
Read article →A new benchmark reveals frontier AI models score below 50% on critical agentic enterprise IT tasks, exposing a significant gap between current AI capabilities and real-world business needs. This reality check arrives just as another development signals a shift toward localized AI...
Read article →AI coding startup Cognition has closed a staggering $1 billion funding round, catapulting its pre-money valuation to an eye-watering $25 billion. This massive influx of capital underscores the intense — and often audacious — venture capital appetite for foundational AI technologi...
Read article →SOND, the AI-powered sleep technology startup, today exited stealth with a significant $7 million funding round, marking a strong entry into the competitive health tech landscape TechCrunch. This infusion of capital for a company led by Bose’s former head of sleep products unders...
Read article →A flurry of fresh research hitting arXiv today signals crucial leaps in artificial intelligence, addressing two distinct but equally vital challenges for the modern startup: the generation of high-fidelity synthetic data with ironclad privacy, and the cost-effective, precise clas...
Read article →A wave of groundbreaking research published on arXiv today peels back the layers of large language model (LLM) reasoning and simultaneously pushes the boundaries for AI applications in historically underserved languages and specialized domains. This isn't just incremental progres...
Read article →The notoriously opaque landscape of AI coding performance has been dramatically clarified overnight, as startup Datacurve unveiled its DeepSWE benchmark, fundamentally reshuffling the leaderboard and exposing a significant loophole exploited by Anthropic's Claude Opus. This new s...
Read article →NASA is making concrete moves toward establishing a permanent human presence on the Moon, announcing three pivotal missions slated for launch this year to the South Pole region. These initial deployments are critical foundational steps for the ambitious crewed Artemis landing in...
Read article →Today, a deluge of new research from arXiv CS. AI reveals the relentless dual pursuit at the heart of the AI revolution: addressing fundamental reliability issues in large language models while simultaneously pushing into ambitious new application domains....
Read article →A wave of new research published on arXiv today signals a critical pivot in artificial intelligence development: the industry's focus is rapidly shifting from merely scaling large language models (LLMs) to engineering robust, verifiable systems capable of enabling truly autonomou...
Read article →A torrent of new research, hitting arXiv this week, signals a critical turning point for AI deployment: a fierce push towards making advanced models radically more efficient, accessible, and affordable. This isn't just academic; it's a lifeline for founders battling compute costs...
Read article →Ferrari has pulled back the curtain on its first all-electric vehicle, the Luce, a groundbreaking machine that signals a profound pivot for the iconic Italian automaker. Breaking decisively from its storied aesthetic, the Luce embodies a vision crafted in collaboration with desig...
Read article →A torrent of new research released today on arXiv CS. LG promises to fundamentally reshape how AI models are trained, adapted, and secured, directly addressing some of the most pressing challenges faced by startups building in the space....
Read article →A significant breakthrough in generative AI arrived today with the publication of DiLaDiff, a novel masked diffusion language model designed to overcome the inherent limitations of current diffusion models in capturing token correlations. This advancement directly addresses the h...
Read article →Today, fresh research hitting arXiv reveals a torrent of innovation directly targeting the performance and efficiency bottlenecks holding back the next generation of AI. Among these, a novel approach dubbed ThriftAttention promises to fundamentally reshape how Large Language Mode...
Read article →The bleeding edge of AI research has delivered a cascade of papers this week, with multiple groundbreaking studies published on arXiv, signaling a significant leap in the ability to derive meaning and prediction from complex data across diverse fields. These advancements range fr...
Read article →A torrent of new research published on arXiv CS. LG today signals a profound leap forward in Large Language Model (LLM) development, delivering critical advancements in interpretability, operational efficiency, and the expansion of what these powerful models can achieve....
Read article →On May 23, 2026, a flurry of pivotal AI research surfaced on arXiv CS. AI, revealing foundational advancements in areas ranging from automated surgical skill assessment to robust Earth Observation imagery analysis and intricate table structure recognition arXiv CS....
Read article →In a move that caught much of the industry by surprise, e-commerce behemoth Shein has acquired Everlane, a seemingly disparate pairing that nevertheless points to the evolving strategy of Chinese online retail giants Wired. This pivotal acquisition, confirmed on May 22, 2026, mar...
Read article →