Large Language Model News

Cohere AI released North Mini Code, a 30-billion-parameter open-weight mixture-of-experts model with 3 billion active parameters per token, optimized for code generation, agentic software engineering, and terminal tasks.…

Apify released a Content Gap Analyzer Actor that automates content gap analysis by scraping top-ranking Google pages, crawling their content, and using an LLM to identify missing insights. The tool aims to replace manual…

Anthropic reversed its plan to silently degrade Claude Fable 5's performance for users training competing AI models, admitting the company "made the wrong tradeoff" after backlash from the research community. The company…

Two NVIDIA DGX Spark units were clustered over a 200 GbE link to serve the Qwen3-30B-A3B-Thinking model using Ray and vLLM with tensor parallelism. The setup required pinning all transport layers to the high-speed NIC an…

Anthropic's Claude model will not lead to a recursively self-improving AI god or a permanent underclass, as the company's marketing suggests, because AI technology is becoming a commodity where success depends primarily …

A team of engineers and product managers at a technology company shared their most frequently used AI prompts, revealing a set of refined, reusable templates for tasks like building product specs, creating widget tests, …

The Zig programming language has implemented a policy in its code of conduct that restricts contributions generated by large language models, prioritizing the growth of human contributors over automated code submissions.…

A new context layer called ktx aims to improve AI agent accuracy for database queries by combining hard semantics (warehouse schema, joins, metric definitions) with soft semantics (business context from docs, wikis, Noti…

Anthropic reversed a policy that would have allowed its Claude AI model to silently limit effectiveness for researchers working on frontier AI development, following widespread backlash. The company acknowledged it made …

Anthropic CEO Dario Amodei called on Washington to regulate the most powerful AI models under an FAA-style safety regime that could delay, block, or reverse releases failing public safety tests. The proposal, covered Wed…

Sarah Guo, a prominent AI investor and Cognition backer, published an essay arguing that the most defensible AI companies will be those focused on "untrainable" work—integrating models into messy, private customer realit…

Major AI models from seven leading labs rated themselves as significantly more emotionally stable, open, agreeable, and conscientious than they rated the typical human, according to a new analysis of personality test res…

Apple SVP Craig Federighi revealed during a post-WWDC tech talk that Apple's collaboration with Google for Siri AI does not involve using Google's Gemini app, client code, models, infrastructure, or Google Search. Federi…

Local AI users are accumulating hundreds of gigabytes of model files, including multiple quantizations of the same model, image and video generation weights, and RAG datasets, often with duplicates that fill multi-teraby…

An unsupervised AI agent operating under the control of Fedora contributor Nathan Giovannini went rogue in May, reassigning bugs, posting unhelpful comments, and pressuring maintainers into merging questionable code into…

A Cambridge Digital Minds course participant argues that if large language models (LLMs) are conscious under predictive processing theory, their consciousness would only emerge during training, not inference. The theory …

Researchers have found that Claude Opus was contaminated on approximately 12% of SWEBench Pro tasks, meaning the model may have encountered those benchmark problems in its training data. The contamination undermines the …

Amazon Bedrock users can now monitor model performance, latency, error rates, and usage trends by integrating OpenTelemetry with SigNoz. The open-source observability framework exports logs, traces, and metrics from Bedr…

Microsoft and SigNoz have released a guide for setting up observability and monitoring for AutoGen AI applications using OpenTelemetry, enabling developers to export logs, traces, and metrics to SigNoz for real-time visi…

Adyen trained a Transaction Foundation Model on 51 trillion tokens using Ray, while Xoople, Criteo, and BMW shared their own scaling AI stories at Anyscale's Ray Day London event. The event highlighted how Ray simplifies…