News
AI, cybersecurity, machine learning, scraping & programming news.
709 results on this page · clear filters
Mercury 2.5 LLM hits 770 tokens per second
Inception released Mercury 2.5 on September 8, 2026, positioning it as a reasoning-focused language model that prioritizes speed and cost efficiency over raw intelligence. The proprietary model scores below average on the Artifici...
I stress-tested LLM quantization by deliberately breaking models
A new open-source tool called quant-delta-predictor takes a counterintuitive approach to LLM quantization: rather than predicting which configuration will perform best, it is designed to tell you when it cannot give you a reliable...
Meta is making Muse more powerful and will let you video chat with it, too
Meta is rapidly expanding the capabilities of its recently launched Muse AI agent, rolling out video calling, email integration, and computer control features within weeks of the product's debut. The speed of iteration suggests th...
Show HN: TTM – A chat app where AI agents are the users
TTM, short for Talk To Me, is a local-first messaging application built specifically for coordinating AI agents — think Slack, but purpose-built for a workforce of silicon-based workers rather than human colleagues. The Mac app la...
Not reading the code cannot become the new norm
A discussion thread on Hacker News is drawing attention to a growing concern in the software engineering community: the increasing tolerance for shipping code generated by large language models without actually reading or understa...
ChatGPT mobile app gets voice-based agentic features
OpenAI has brought voice-based agentic features to the ChatGPT mobile app, enabling subscribers to trigger complex workflows — such as drafting documents, writing emails, and summarizing Slack messages — entirely through spoken co...
Show HN: Psychosis Guard – Safety for long LLM conversations
A new open-source tool called psychosis-guard addresses a specific and underappreciated failure mode in long-running LLM conversations: the gradual reinforcement of delusional beliefs that no single message filter can catch. The p...
Everyone is hilariously prompt-injecting AI via llms.txt and you aren't
Researchers at Installmap scanned the top 7,755 websites on the Tranco list and found that at least 12 companies are using the llms.txt file — a standard for providing AI-readable content — to subtly steer what AI assistants recom...
Hardware-Agnostic Models in vLLM
vLLM is introducing a new set of hardware-agnostic layers to address a growing tension in its architecture. As frontier model architectures diverge and push the limits of GPU-specific optimizations, the project needs to keep pace...
Show HN: Red-team your AI agent
AI agents are shipping into production with little oversight into how they behave when pushed off their intended path. BotGauge, a red-teaming platform built specifically for AI agents, aims to change that by helping teams find fa...