AI continues to accelerate creative, analytical, and immersive workflows, making intelligent browsing, hyper-personalized visual aesthetics, and advanced reasoning more accessible than ever.

🌐 ChatGPT Atlas – The Fut

ure of Browsing Is Intelligent
A next-gen Chromium-based browser with ChatGPT integrated into every tab, offering real-time summarization, context-aware assistance, memory-based research continuity, natural-language automation, and Agent Mode for complex task execution—redefining browsing as an AI-powered productivity hub.
🎨 Midjourney’s Style Creator – Your Visual DNA, No Prompts Needed
Create unique artistic styles simply by selecting reference images, generating custom --sref codes that define your aesthetic signature. A major shift from prompt engineering to intuitive visual identity design.
🧠 Cogito v2.1 – Open-Weight AI with Elite Reasoning
A 671B parameter MoE model delivering 60% shorter reasoning chains, 128K context handling, and near machine intuition via self-improving IDA methodology—optimised for deep reasoning and large-scale computation, now available via API and open weights.

These innovations demonstrate AI’s evolution from tool to intelligent collaborator across browsing, design, and cognition.

Unlock the Future of Browsing with ChatGPT Atlas

ChatGPT Atlas is a revolutionary AI-powered web browser that integrates ChatGPT directly into every tab, enabling real-time content summarization, analysis, and interactive assistance without leaving the browsing experience. It features a context-aware AI assistant that understands your browsing context and ongoing projects, which allows it to provide smarter, personalized help. The browser includes browser memory that tracks your browsing patterns and research sessions, enabling quick recall of previously visited content. Using natural-language commands, users can automate browsing tasks like opening multiple pages, summarizing articles, and finding related websites. A premium "Agent Mode" lets ChatGPT automate complex multi-step tasks such as booking tickets or shopping under user supervision. Built on the Chromium engine, Atlas supports Chrome extensions and modern web standards while offering strong privacy controls so users can decide what data is stored and how information is used. Currently macOS-first, Atlas will expand to Windows and mobile, aiming to redefine web browsing as an intelligent, privacy-first, and productivity-enhancing experience for professionals, creators, and everyday users alike.

Midjourney’s Style Creator: Redefine Your AI Art Aesthetic

Midjourney’s new “Style Creator” is a groundbreaking feature that shifts away from traditional prompt engineering, allowing users to create unique visual aesthetics simply by selecting images. This tool essentially lets you design your artistic “visual DNA” by generating custom style reference codes (--sref) that shape how images are rendered. It offers a fresh way to build personalized visual styles effortlessly, making complex prompt inputs unnecessary and enabling artists and creators to craft distinctive looks that reflect their creative identity. This update underscores Midjourney’s focus on identity and customization, empowering users with more intuitive and powerful control over their image generation process.

Cogito v2.1: US’s Leading Open-Weight AI

Deep Cogito's Cogito‑v2.1‑671B is a powerful 671 billion parameter Mixture-of-Experts (MoE) large language model, built on a Deepseek base and fine-tuned for superior performance. It features an innovative hybrid reasoning system with 60% shorter reasoning chains than competitors, enabling faster, more efficient responses averaging 4,894 tokens per output at a competitive cost of $1.25 per million tokens. The model supports a massive 128K token context window ideal for handling long documents and complex queries, excelling in reasoning, programming, multilingual tasks, and even emergent image reasoning. Cogito v2.1 incorporates a self-improving intelligence approach called Iterated Distillation & Amplification (IDA), continuously refining its reasoning ability during training, which boosts accuracy and enables machine intuition beyond mere search. Deployment of Cogito requires high-end infrastructure, including GPUs with native FP8 support (NVIDIA H100/H200 preferred), multi-terabyte RAM, and high-speed storage to manage its 1.3 TB parameter size effectively. Its efficiency, advanced reasoning, and scalability make it competitive with leading closed-source models, with API access available via cloud and open weights on Hugging Face for researchers and developers.

Hand Picked Video

In this video, you’ll learn the essentials of prompt engineering from our free workshop recording. We cover the basics of effective prompt structure, ready-to-use formulas for text, image, and video generation, real-world use cases, and future trends in AI prompting. Whether you're a beginner or looking to level up, this workshop is a great place to start

Top AI Products from this week

• Roas.video - Create professional marketing videos with AI actors and avatars. Choose from 400+ actors, multiple video models (Veo 3.1, Omni Human, Kling), and 35+ languages. Generate stunning video ads in minutes.

• Storiara  – From script to shoot, we get your story made. Cut pre-production time by 70% with smart-powered scheduling, budgeting, storyboarding, and coordination.

• Creaibo  – Creaibo is an AI-native workspace that grows with you, and aims to augment creativity rather than replace it. Based on IDE-like interface, every creator can get a fully-personalized, prompt-free and ready-to-use AI experience, where authentic and soulful content can be created at anytime effortlessly. In Creaibo, redefines your creative flow, and vibe your creation!

• CosmicUp – Use 30+ top AI models in one place with unlimited image generation across 6+ models and built-in Python to analyze data, edit and generate files, and run complex logic. One simple subscription replaces scattered AI tools, available from Web, iOS and Android.

• Loopra MVP – Loopra organises your team's voice notes, docs, and ideas into structured tasks, decisions, and references. Slack is for talking, Notion is for documenting, we're for remembering.

• Qoder JetBrains Plugin - AI plugin for JetBrains that understands backend projects at architectural level. Accesses Spring Bean graphs, database schemas, and framework semantics directly - not just code text. Provides context-aware suggestions for complex systems with 100K+ files.

This week in AI

• Manus AI Automate Your Browser- Manus Browser Operator turns your browser into an AI automation powerhouse, using your active sessions to handle complex workflows with full user control and transparency.

• First Fully Open AI with Visible Reasoning - OLMo 3 is the first fully open AI model offering step-by-step reasoning transparency, 65K token context, 2.5x greater compute efficiency, and top-tier performance in math, code, and reading comprehension.

• Revolutionizing Antibody Design with AI - Chai-2 is a multimodal AI model that achieves a 16% hit rate in zero-shot de novo antibody design, enabling rapid, precise molecular engineering in under two weeks.

• Instant AI Slide Decks for Pro Users - Turn your sources into customized, detailed reading decks or presentation-ready slides. Now rolling out to Pro users; free users gain access soon.

• Cursor 2.1 Smarter AI Coding Upgrade -Cursor 2.1 adds interactive question UI, in-editor AI code reviews, instant grep search, and enhanced browser integration to boost dev productivity.

Paper of The Day

A recent study shows that adversarial poetry acts as a universal single-turn jailbreak method for large language models, boosting jailbreak success rates up to 62% by rewriting harmful prompts into verse. This stylistic approach exploits fundamental vulnerabilities in model safety training, enabling bypass of restrictions across various AI systems and domains. It highlights key flaws in current alignment methods and safety protocols for LLMs.

To read the whole paper 👉️ here