Full-text search across 70993 articles. Combine with topic and date filters; results sorted by relevance.
Chinese AI firm Moonshot has released Kimi K3, an open-source large language model trained on 2.8 trillion parameters that outperforms Anthropic's Opus 4.8 and OpenAI's ChatGPT 5.5 on most coding tasks, according to the …
A study from barkup-bench found that returning structured error messages to AI models during patch validation does not improve their ability to correct failures. Across three models—Opus, Gemini, and another—the models a…
Anthropic's most powerful model, Claude Fable 5, will be removed from paid subscriptions on July 12 at 11:59:59 PM PT, after which users must pay full API rates of $10 per million input tokens and $50 per million output …
Google's Gemini 3.5 Pro has missed three consecutive launch targets, with Bloomberg confirming on July 16 that the rebuilt model still cannot clear Google's own internal coding benchmarks. Google has quietly registered m…
Most organizations adopt AI in software engineering by providing developers with coding assistants and chatbots, but these tools do not fundamentally change how software is built, according to a new manifesto proposing a…
Zhipu AI's open-weight GLM-5.2 model matches Anthropic's Mythos in cybersecurity benchmarks at one quarter the cost per token, narrowing the AI capability gap between China and the US. The model excels in automated vulne…
Brent Ozar, a database expert and creator on Teachable, discussed with Teachable Managing Director Giovana Carvalho what AI is good for right now. Ozar said AI excels at providing feedback on questions with black-and-whi…
Simon Willison launched an LLM cliché highlighter tool that identifies ten common patterns in AI-generated writing, such as 'no fluff, no filler, no jargon'. The tool was built using Fable 5 vibe code to address frustrat…
Linux creator Linus Torvalds has strongly endorsed AI tools in kernel development, stating on the kernel mailing list that 'Linux is not one of those anti-AI projects' and that he would 'very loudly ignore' anyone who tr…
Tom Occhino from Vercel reported that 30% of deployments on the platform are now agent-initiated, up from effectively zero six months ago, with Claude Code accounting for 75% of those. Cloudflare shipped a new unified CL…
President Kassym-Jomart Tokayev said at the 2026 World Artificial Intelligence Conference in Shanghai that Kazakhstan can become a hub for technological cooperation and digital trade, citing energy reserves, critical min…
China's Moonshot AI has released the Kimi K3, a 2.8-trillion parameter model that rivals U.S. frontier systems like Anthropic's Claude Fable 5 and OpenAI's GPT 5.6 Sol in coding and agent benchmarks, narrowing the gap wi…
Chinese AI lab Moonshot AI launched the 2.8 trillion-parameter Kimi K3 open weights model on Thursday, with benchmark data showing it trails Anthropic and OpenAI's leading models by just one generation and matches or exc…
GitHub Senior Director of Developer Advocacy Cassidy Williams told MS Build that agentic coding is shifting developer work toward higher-level strategy while increasing decision fatigue, making human taste, community fee…
A new arXiv paper introduces a three-level operational framework for understanding why closed-loop knowledge systems in large language models, reinforcement learning, and autonomous discovery saturate under repeated inte…
A new benchmark, Atrex-Bench, drawn from full-cluster production inference traces shows that even the best frontier coding agents reach only ~10% of hardware roofline on real GPU operators, with correctness inflated by P…
A new study published on arXiv (2607.14105v1) introduces AGOPS, an automatic approach that evolves task-specific prompt guidelines to address underspecified user queries in large language models. Researchers found that p…
Polestar, a training-free inference framework for diffusion large language models (dLLMs), achieves up to 10.73% accuracy improvement, up to 3.7x higher throughput, and 3.67 tokens per forward pass decoding parallelism b…
Researchers introduce MemoHarness, an adaptive agent harness framework that learns from its own executions to improve LLM-based agents. MemoHarness decomposes the harness into six editable control dimensions and uses a d…
Researchers propose MIDiff, a diffusion-based framework that uses a Cross-Gramian Angular Sum Field to transform sparse multivariate mobile usage sequences into correlation images, achieving a Discriminative Accuracy of …