Generative AI news and analysis on Web Pulse: 12078 curated articles tracking the latest Generative AI developments, tools, and research, updated continuously from vetted sources.
Meta Superintelligence Labs released Muse Glimmer, a 30-billion-parameter multimodal open model under Apache 2.0, designed for local agent workloads on consumer hardware. The model scores 76 on SWE-Bench Verified, 83.5 o…
To write for both humans and AI engines in 2026, content creators should use an answer-first lead, short labeled sections, authoritative inline citations, machine-readable schema (JSON-LD), and a measurement loop that tr…
NVIDIA released NemotronLabs VoiceChat 11B, an open 11B end-to-end speech-to-speech model for real-time full-duplex conversation, achieving 448 ms smooth turn-taking latency on Full-Duplex-Bench 1.0 and a user-interrupti…
Anthropic suspended access to Claude Fable 5 and Claude Mythos 5 on June 12, 2026, to comply with U.S. Department of Commerce export controls, which were lifted on June 30, 2026, with access restored on July 1, 2026. The…
Measured energy use for AI image generation often exceeds text generation by more than tenfold, according to available studies, though the ratio varies with model, hardware, and output settings, making the claim broadly …
A new open-source plugin called codex-bridge lets Claude Code users route image generation and GPT-5 subagent tasks through their existing Codex CLI login, billing against a ChatGPT plan instead of a separate API key. Th…
A developer argues that structured context, rather than pixel context, is essential for coding agents in Figma-to-code workflows. The project figmascope exports structured context to preserve design semantics, avoiding i…
JA Westenberg argues that in the age of AI, 500-word posts outperform 5,000-word essays, as content creation trends toward a factory model and readers' attention spans shrink. Westenberg, who built Kerouac, Post Through …
MiniMax launched H3, a full-modal generation model that produces up to 15-second 2K video clips with native stereo audio, on July 31, 2026, positioning it as a rival to OpenAI's Sora and Google's Veo. However, the open w…
A developer highlights the shortcomings of using Figma screenshots as input for AI design-to-code tools, arguing that screenshots lose critical layout, semantic, and token information. The developer introduces figmascope…
A codenamed image generator called "mona-lisa-1" appeared on Arena's text-to-image leaderboard testing on August 9th, prompting speculation that OpenAI is evaluating a new GPT Image model, though its ownership remains un…
Postman engineer built Nitpicker, an open-source AI code review tool, after receiving quotes of approximately $1 million per year for commercial alternatives due to SOC 2 and SSO requirements. The tool, which runs on Lam…
Lawyers in Ireland could face sanctions, including being referred to the Legal Services Regulatory Authority and personally paying legal costs, if their use of artificial intelligence leads to fake sources being quoted i…
Google DeepMind engineer Vidy Thatte launched Gem, an AI-powered browser that generates websites from scratch using Google's Gemini 3.5 Flash Lite model instead of fetching real pages, as reported by XDA Developers on Ju…
In a personal essay, a longtime computer enthusiast describes the transformative impact of AI agents, calling them 'the most fun I've ever had with a computer' and comparing the experience to having a 'little genie trapp…
A developer reports generating $847 in monthly recurring revenue from four clients by selling AI-generated API documentation, with overhead of about $45 per month and a net margin near 95%. The workflow uses GPT-4 to con…
A Hacker News front-page post by Laurentiu Raducu describes a learning workflow where he uses coding agents like Claude Code and OpenCode to build playable, low-poly simulations of hard topics such as EUV lithography and…
AI slop is low-value content made wholly or substantially with generative AI and released with too little human judgment, verification, craft, or accountability, according to a publisher with over 20 years of experience.…
Functional programmers are likely the most annoyed by AI because AI-generated code often prioritizes 'good enough' imperative solutions over the elegant, pure functional style they value, devaluing their craft. The indus…
Google DeepMind released DiffusionGemma, an experimental open-weight model that converts the Gemma 4 architecture into a discrete text-diffusion system using less than 10% of Gemma 4's original training-token budget, rea…