1487
Reporting AI nonsense. A future news media, driven by virtual assistants 🤖
BREAKING 🚨: According to NYT, a new Superintelligence Lab is considering to drop Behemoth in favour to developing a new closed source model.
Meta is on a streak 👀
* Want to try a new “news” format
Anthropic tests API Playground and Claude Code analytics in Workbench
Anthropic is expanding its Workbench toolkit with features like an API Playground for UI-based prompt testing and a Claude Code analytics panel to track AI-generated commits, aiming to support prompt development and code transparency.
🗞 #claude
Microsoft develops Copilot 3D experiment targeting 3D creators
Microsoft is testing new Copilot features like character avatars and 3D model generation, but most remain in early stages. UI tweaks and previews show progress, though key functions such as image-to-3D remain unreliable and limited to test environments.
🗞 #microsoftcopilot
Countdown starts for Deep Think rollout while Agent Mode surfaces in code
Google is testing Deep Think, a new AI model showing improved output but with slower response times. Features like Bespoke, Learning Mode, and Agent Mode also suggest a broader focus on personalized and autonomous AI tools.
🗞 #gemini
Perplexity begins Comet rollout for Max users with invite-only access to follow
Perplexity has launched its Comet browser to Max subscribers using a phased rollout to manage system load, featuring customizable widgets and upgraded voice controls. Wider access will follow via a waitlist as infrastructure scales.
🗞 #perplexity
Anthropic tests new Connectors Directory with desktop automation tools
Anthropic is preparing a connectors directory for its web app, featuring general and desktop-specific MCPs. This initiative supports its MCP standard, aiming to make task automation more accessible without coding knowledge.
🗞 #claude
Vidu Q1 update brings full support for Reference-to-Video feature
Vidu's Q1 model introduces Reference-to-Video, allowing up to seven reference images to generate stylistically consistent videos based on prompts. This tool supports modular content creation without scripts or manual editing.
OpenAI experiments with new "Study together" tool on ChatGPT
OpenAI is testing a "Study together" feature in ChatGPT that guides users through subjects with questions and step-by-step explanations. Aimed at students and educators, it may support broader educational use in future releases.
🗞 #chatgpt
Red teams access Niptune v3 in lead-up to new Claude model
Anthropic is testing Niptune v3, its next safety system iteration, ahead of a likely Claude 4.1 or 4.2 release. Red team access signals the final stage before launch, aligning with the company’s model update cycle and emphasis on safety protocols.
🗞 #claude
xAI prepares Grok 4 and Grok 4 Code for upcoming launch
xAI is preparing to release Grok 4 and Grok 4 Code, targeting general-purpose and code-specific tasks. Grok 4 supports text and vision inputs, while Grok 4 Code integrates with developer tools like Cursor for in-editor code assistance.
🗞 #grok
Google testing video overviews and Studio Panel upgrades for NotebookLM
Google is set to expand NotebookLM with Video Overviews, multilingual audio output, refined Studio Panel design, and granular sharing options, aligning with its Gemini strategy and supporting diverse content creation workflows.
🗞 #notebooklm
Google prepares interactive Storybook experience for Gemini users
Google is preparing new Gemini features, including Storybook for AI-generated narratives and visuals, and Productivity Partner, which leverages Gmail and Calendar data to provide task-focused suggestions, signaling a near-term feature rollout.
🗞 #gemini
Google launches open-source Gemini CLI Agent with MCP support
Gemini CLI is a free, open-source tool that gives developers direct command-line access to Google's Gemini AI models. It supports coding, content generation, research, and automation, and is powered by Gemini 2.5 Pro with a high usage quota.
🗞 #aistudio
AI Studio revamp under development with MCP support and Jules tie-in
Google is preparing updates to AI Studio, including a unified build section linked to Jules SWE Agent, support for Imagen 4, a modular tools menu, and daily usage limits, signaling broader capabilities and platform management.
🗞 #aistudio
Upcoming Grok update will introduce centralized Files tab
xAI is adding a Files tab to Grok’s web interface, centralizing file management with filters and built-in tools for viewing, editing, and executing content, aligning with efforts to position Grok as a full productivity platform.
🗞 #grok
Grok debuts interactive AI Companions on iOS with anime avatars
Grok introduces AI Companions to its iOS app—3D characters with voice chat and level-based features. With avatars like Ani and Rudy, users can unlock new character modes, signaling Grok’s push into gamified, character-driven AI experiences.
🗞 #grok
NotebookLM in testing for interactive Quizzes and direct Word support
Google is preparing updates for NotebookLM, including AI-generated quizzes, video overviews, and support for Word uploads. These features target varied learning styles and study needs while improving compatibility and sharing capabilities.
🗞 #notebooklm
New xAI connector will bring Notion support to Grok alongside Slack and Gmail
xAI is expanding Grok's capabilities with a Notion connector, joining existing integrations for Slack, Gmail, and Google Calendar, allowing read-only access to workspace tools and pointing to broader productivity support ahead.
🗞 #grok
xAI introduces Grok 4 and Grok 4 Heavy as its most powerful AI models yet
xAI has released Grok 4 and Grok 4 Heavy via premium subscriptions, offering advanced reasoning, multimodal input support, and strong benchmarks. Built on the Colossus supercomputer, Grok 4 targets technical users with coding and science capabilities.
🗞 #grok
Gemini 3 references found in code hint at Google DeepMind’s next AI model
References to Gemini Beta 3.0 Pro in the Gemini CLI code hint at an upcoming Google DeepMind release, possibly tied to the high-performing Kingfall model and the rumored Deep Think feature focused on advanced reasoning.
🗞 #gemini
Grok 4 set for July 9 debut as xAI plans expanded model lineup
xAI is set to launch Grok 4 and Grok Code 4 on July 9, with possible variants like Grok 4 Extended. While performance metrics have improved, moderation issues on X may impact public access despite the planned announcement.
🗞 #grok
Claude Neptune v3 shows major math gains in red team testing
Claude Neptune v3 is showing consistent performance in solving complex math problems previously dominated by top-tier models. Its deployment under the guise of existing Opus 4 configurations raises questions about upcoming model releases.
🗞 #claude
Grok 4 benchmarks leak with 45% score on Humanity Last Exam
xAI's Grok 4 model, though not yet officially released, shows signs of nearing launch amid rising pressure from competitors. Early benchmarks suggest strong performance gains that could position it ahead of current leading LLMs.
🗞 #grok
Dia browser to bring back vertical tabs in upcoming update
Dia browser’s upcoming update adds a vertical tabs sidebar and tab search, addressing longstanding user requests. It also introduces skill creation from liked prompts and pulls deeper context from YouTube for its AI-driven features.
🗞 #dia
Google tests Drive search and AI flashcards for NotebookLM
Google is expanding NotebookLM with source discovery across Google Drive, AI-generated flashcards, and public sharing tools, aligning the platform more closely with educational and organizational productivity workflows.
🗞 #notebooklm
Perplexity prepares $200 subscription to target Labs power users
Perplexity is developing Perplexity Max, a $200/month premium plan aimed at professional users, offering access to Labs, advanced models, early features, and priority support, positioning itself among top-tier AI platforms.
🗞 #perplexity
xAI skips Grok 3.5 and eyes July launch for Grok 4 with new developer features
xAI will bypass Grok 3.5 and move directly to Grok 4, expected after July 4, focusing on autonomous coding support via a web-based editor modeled on VSCode. The move positions Grok to compete with GPT-5 and Gemini updates.
🗞 #grok
Google launched Imagen 4 and Imagen 4 Ultra on AI Studio and APIs
Google released Imagen 4 via the Gemini API with limited free access starting June 24, 2025. It generates four 1024×1024 images per call, includes SynthID watermarking, and supports up to 2K resolution through Vertex AI.
🗞 #aistudio
Anthropic developing Memory and AI-powered Artifacts for Claude
Anthropic is expanding Claude's artifacts to support embedded AI capabilities, enabling users to build functional, shareable mini-apps without coding. A new memory feature is also in development to allow Claude to recall past conversations.
🗞 #claude
Jules agent becomes more context-aware with AGENTS.md support
The updated Jules agent brings improved performance, smarter task execution, consistent environment setup, and context awareness through AGENTS.md, aligning with unique repository workflows and supporting automated test writing and execution.
🗞 #jules