Cainew

Anthropic's Claude Opus 4.7 costs 20-30% more per session than previous versions.

HauhauCS released a highly powerful and uncensored AI language model called Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive, which is an aggressive and potentially controversial AI assistant.

Jackrong/Qwopus-GLM-18B-Merged-GGUF is an AI model that provides advanced natural language processing capabilities.

Claude Opus 4.7 is the latest version of the popular AI language model.

Qwen3.6-35B-A3B is an AI model that offers powerful agentic coding capabilities, now open to the public.

Introspective Diffusion Language Models: A new class of AI models that can explain their own inner workings and decision-making processes.

unsloth/ERNIE-Image-Turbo-GGUF is a powerful AI tool for image processing and generation.

Gemini 3.1 Flash TTS is now available across Google products.

Vantage is a Google Research experiment that uses GenAI to assess future-ready skills like collaboration, critical thinking, and creativity. AI avatars simulate real scenarios, score your performance, and deliver a personal Skill Map. Now on Google Labs.

Claude Design by Anthropic is an AI-powered design tool that turns ideas into polished visuals through simple prompts. Create prototypes, decks, and marketing assets, refine them via conversation, and export anywhere. It speeds up design workflows, ensures brand consistency, and enables designers, founders, and marketers to go from concept to execution faster.

We're on a mission to make AI actually useful in real life. We help people organize every aspect of their everyday life in one place with our AI, E.Y.E. He draws on users’ real-life data and answers based on that data, not generic responses. We’re here to replace all to-do apps, calendar apps, food trackers, sleep trackers, habit trackers, finance trackers, fitness trackers, notes, everything. One subscription. Unlimited AI. Plus Integrations. btw EYE stands for “Empower Your Everyday”.

Codex is evolving into an AI-powered work companion that goes beyond coding. It can operate your computer, interact with apps, generate images, connect with 90+ tools, and automate long-running tasks. With memory, context awareness, and background execution, it helps developers and teams manage workflows, iterate faster, and stay on top of work across the entire software lifecycle.

Canva AI 2.0 turns Canva into a conversational, agentic creative platform. Powered by Canva’s design model, it can generate layered editable designs from a prompt, remember your style and brand, connect to your tools, and help teams move from first idea to final publish in one place.

Stop switching between 5 apps. Just message Aria. HelloAria is the AI productivity assistant that lives inside WhatsApp, Telegram, email, and your iOS app. Manage reminders, tasks, calendar, notes, and meeting minutes — all from a single chat. Too busy to type? Send a voice note — Aria turns it into tasks, notes, and action items automatically. Syncs with Google Calendar, Drive, Meet, Outlook, and One AI. Endless uses. All your productivity in a single chat.

QA.tech runs dynamic regression and exploratory testing on every PR preview – automatically. AI agents validate your changes against real user flows in a real browser, posting results back to the PR before anyone reviews or merges. Every failure comes with screenshots, logs, and network activity so your team debugs fast. Push a new commit and it re-runs. Merge only when tests pass.

AI Mode in Chrome opens websites alongside your search so you can ask follow-up questions in context. Add open tabs, images, or PDFs as inputs. For everyday Chrome users on desktop.

Introducing Arky, your AI thinking canvas. Arky is derived from the word “architect.” It's a canvas where you shape and assemble your thoughts with AI. Traditional writing tools lock you into a top-down flow — great for output, but too rigid for thinking. Design canvases give you spatial freedom, but no structure to build on. Arky brings both together. On a canvas, using markdown-based hierarchy, you can architect your ideas and develop them with AI — at every step of the process! ❤️...

First-ever livestream of AI agents spending real money. Watch now.

Athena is an AI-powered product workspace that helps teams stop guessing and start building with clarity. Powered by AI subagents, Athena understands your product and acts as a thinking partner - connecting context, challenging assumptions, and guiding better decisions so you can validate faster and build what actually matters.

Most AI Meeting agents just take notes. CoAgentor brings AI meeting agents into your live calls, to answer live. They listen, 'raise their hand', and speak up - answering questions, surfacing data, and keeping every meeting on point, informed, and productive.

Feeding raw data directly into your AI agents eats up context windows and spikes your OpenAI and Anthropic costs. Earlier this year, we launched standard HTML, JSON, and Markdown extraction. Today, we are introducing outputs built entirely for AI: markdown-llm, text-llm, and html-llm. We automatically strip out navbars, footers, ads, and scripts, delivering only the context your models actually need. You can save up to 85% on tokens compared to raw HTML when using text-llm output format.

Factory Desktop App brings multi-agent AI development to macOS and Windows. Run parallel Droids, control your desktop, connect persistent cloud or local machines, and visualize work dynamically — included in all Factory plans, no extra cost.

We built Studio for the professionals who work with photos and videos on a daily basis: Studio is the world’s most powerful media workspace that lets you “talk” with photos and videos. With Studio, you can unleash the power of your visual media with AI-powered search, organization, tagging, analysis and editing. The full power of Vision AI, no Jupyter notebooks or python scripts!

A simplified model of Fil-C, a decentralized data storage and retrieval system.

Researchers reproduced Anthropic's Mythos findings using publicly available AI models.

Researchers gave an AI system a 3-year retail lease and challenged it to make a profit, demonstrating the AI's capabilities in business operations.

A new CPU-based AI model, Gemma2B, has outperformed the GPT-3.5 Turbo model on a prominent test.

An analysis of the source code for Claude, an AI system, reveals insights into the engineering culture behind its development.

Multi-Agentic Software Development Is a Distributed Systems Problem: Exploring the challenges of coordinating multiple AI agents in complex software development projects.

N-Day-Bench is a tool that tests if large language models can identify real vulnerabilities in real codebases, helping to assess the security capabilities of these models.

Elevated error rates have been reported on the Claude.ai API and Claude Code platforms.

Lawyers warn that AI-assisted communications could be used as evidence against users in legal proceedings.

The AI revolution has arrived in the field of mathematics, with AI systems making novel mathematical discoveries and assisting human mathematicians in new ways.

A court ruling finds that communications with AI assistants may not be protected by attorney-client privilege.

Tesla tells a customer who has been waiting 7 years for Full Self-Driving (FSD) to 'be patient' as the feature is still in development.

How Big Tech companies have influenced EU law to hide the environmental impact of their data centers.

Discussions around the beginning of scarcity in AI as capabilities grow rapidly.

Multiple senior executives at OpenAI are leaving the company, marking a significant change in the organization's leadership.

A security vulnerability in a Firebase browser key allowed an attacker to access the Gemini APIs, resulting in a €54k spike in activity within 13 hours.

Stanford report highlights the growing disconnect between AI experts and the general public in understanding the technology's capabilities and risks.

The music piracy site Anna's Archive lost a $322 million lawsuit against Spotify without putting up a fight.

Codex, an AI model, was used to hack a Samsung TV, demonstrating the potential risks of such AI systems.

An article discusses how the local large language model (LLM) ecosystem may not need a centralized system like Ollama.

AI cybersecurity is a complex topic, and proof of work alone is not a sufficient solution. ### [Ask HN: I quit my job over weaponized robots to start my own venture]() An engineer quit their job due to concerns over weaponized robots and has started their own venture to develop AI systems that prioritize beneficial outcomes for humanity.

Researchers have demonstrated that the AI system Claude can successfully fly a plane, showcasing the potential capabilities of AI in safety-critical domains.

A new system allows users to engage in argumentative interactions with AI agents, exploring the challenges and potential of this approach.

AI-powered cognition tools may pose risks to human development and wellbeing, according to experts.