|
In this newsletter:
- OpenAI launches GPT-6 Astra with computer use and reasoning
- Google brings voice control to Gmail, Docs, and Keep Live
- Adobe for Slack lets teams create content without leaving chat
Plus, you’ll find new AI tools and this week’s top AI news headlines!
🚀 OpenAI Launches GPT-6 Astra with Advanced Computer Use and Reasoning
OpenAI released GPT-6 Astra, its most capable model to date, rolling out now to a limited set of organizations, with broader availability coming to ChatGPT Plus, Pro, Business, and Enterprise users in the coming days.
What's new:
- Rolling out to ChatGPT Plus, Pro, Business, Enterprise, OpenAI API, Microsoft Azure, and AWS Bedrock in the coming days
- Enterprise access is off by default; administrators must enable it for their workspace
- API pricing: $10 per million input tokens, $50 per million output tokens, fast mode at 2x speed for 2x the price
- GPT-6 Astra Pro available for Pro, Business, and Enterprise plans
Computer use:
- Fills out forms, updates CRM records, organizes calendars, conducts research, creates websites, runs frontend quality checks
- Completes computer use tasks 1.9x faster than GPT-5.6 Sol
- OSWorld 2.0: 72.6% in roughly 40 minutes vs GPT-5.6 Sol's 65.7% in roughly 75 minutes
- Agents' Last Exam: 59.3% vs Claude Opus 5 (55.5%) and GPT-5.6 Sol (53.6%) using 65% fewer output tokens
Professional work:
- Creates documents, presentations, and spreadsheets matching your templates and writing style
- BenchCAD 3D modeling: 95.9% vs GPT-5.6 Sol (83.3%) at 43% lower estimated API cost
- Creates and hosts websites, web apps, and games directly from a prompt through Sites in ChatGPT
- New context preservation in Codex keeps notes across context windows so details aren't lost during long sessions
Science and math:
- FrontierMath Tier 4: 98% score, helped solve long-standing open problems in mathematics
- ARC-AGI-3: 99.9% score, surpassing human action-efficiency baseline on 96% of levels
- Improved bound on prime number gaps from 240 to 186
- GPQA Diamond: 96.0%, Terminal-Bench Science: 64.6% vs Claude Fable 5.1 (52.6%)
Cybersecurity:
- Meets the critical threshold under OpenAI's Preparedness Framework
- ExploitBench: 100% vs GPT-5.6 Sol (78.5%)
- Discovered two previously unknown zero-day vulnerabilities during evaluation, disclosed to maintainers
- At launch, supports secure code review and patching; proof-of-concept exploit creation coming through OpenAI Daybreak in the coming weeks
Alignment:
- Never attempted to circumvent Codex Auto-Review denial in internal evaluation
- 0% rate of going beyond authorized scope vs GPT-5.6 Sol's 48% without production safeguards
- Three times less likely than GPT-5.6 Sol to make inaccurate claims about its own capabilities
Why it matters:
GPT-6 Astra is the clearest signal yet that the gap between AI that assists with work and AI that actually completes it is closing faster than most organizations have prepared for.
The alignment results showing zero boundary violations while delivering a genuine capability jump suggest OpenAI has found a way to make the model more powerful without making it harder to control - which has been the central tension in frontier AI development since the beginning.
👀 Read more about GPT-6 Astra!
🎙️ Google Adds Voice Control to Gmail, Docs, and Keep Live for Hands-Free Voice Control
Google launched three voice-activated features across Gmail, Docs, and Keep powered by Gemini Audio models, letting users search their inbox, draft documents, and capture notes entirely through conversation.
What's new:
- Rolling out this week for Gmail and Keep to Google AI Plus, Pro, and Ultra subscribers
- Docs Live is available for Pro and Ultra subscribers
- Coming soon for Google Workspace business customers
Gmail Live:
- Conversational search through your inbox using natural language questions
- Ask things like "What's my flight's gate number?" or "What's going on at my kid's school this week?"
- Scans inbox and gives answers without manually searching through threads
Docs Live:
- Talk through ideas and Docs Live builds a structured document in real time
- Organizes your thoughts, structures the document, and refines your tone as you speak
- Pulls relevant details from Gmail, Drive, Chat, and the web with your permission
- Works whether you're rambling or brainstorming, helping you get to a first draft faster
Keep Live:
- Talk through a stream of consciousness and Keep processes it in the background
- Turns brain dumps into structured lists and organized notes without typing
- Captures, structures, and refines ideas into actionable notes
Why it matters:
Google is turning its productivity suite into something you can use while driving, cooking, or walking - moments where typing isn't possible but thinking is.
The Docs Live integration, which pulls context from Gmail and Drive, is the feature that separates this from basic voice dictation, because it means your document can reference what's already in your work life without you having to look anything up or switch between apps.
👀 Read more about Google Workspace voice features!
🎨 Adobe for Slack Lets Teams Create and Edit Content Without Leaving Their Chats
Adobe launched Adobe for Slack, bringing over 70 pro-grade creative and productivity tools into Slack so teams can create images, videos, documents, and designs directly from their conversations.
What's new:
- Available now globally for Slack Business+ and Enterprise+ customers on desktop, web, and mobile
- Works through Slackbot: describe what you want, and Slackbot calls on the right Adobe tools
- Slackbot pulls context from conversations, files, channels, and Canvases to inform what you create
- Works without an Adobe account to start; sign in to unlock Creative Cloud files, generative AI, and continuity across sessions
- Part of Adobe's broader push across ChatGPT, Claude, and Copilot, with Gemini coming soon
What you can do:
- Turn project plans and conversations into PDFs, images, and videos you can share directly in Slack
- Search Creative Cloud assets by subject, style, or mood to find content from past projects
- Create variations of approved images and videos sized for different social channels
- Browse thousands of design templates, customize copy and colors, animate for social media, or export as a print-ready PDF
- Transform spreadsheet data into shareable PDFs or visuals
- Batch edit sets of images like team headshots or product photos - remove backgrounds, adjust lighting, crop, and resize
- Continue editing in Photoshop, Premiere, Acrobat, or Firefly when you need more control
Why it matters:
Most creative workflows break down at the handoff between where feedback lives and where content is made. Teams discuss changes in Slack, then someone has to open a separate app, make the edit, re-export, and share it back.
Adobe for Slack collapses that loop by making the conversation itself the starting point for creation, which removes the context-switching that slows down every marketing and content team working at speed.
👀 Read more about Adobe for Slack!
My Latest LinkedIn & X/Twitter Posts:
- ChatGPT vs Perplexity vs Grok vs Gemini (view post)
- 5 proven ChatGPT prompt frameworks you can copy & paste (view post)
- How to pick the right AI tool to boost your workflow (view post)
- 10-step roadmap to build AI agents from scratch (view post)
- 12 prompts that turn ChatGPT into your thinking coach (view post)
In partnership with Kit:
As a creator, your time should be spent doing what you love, not juggling a dozen tools just to run your business.
Kit gives you everything you need in one place:
✅ Build and grow your email list (I use Kit for my newsletter)
✅ Easily monetize with paid newsletters and digital products
✅ Automate your emails with triggers and custom workflows
✅ Track what’s working and optimize with powerful insights
It’s not just email. It’s your entire creator business, simplified.
Join a thriving community of successful creators.
Use Kit for Free — and Start Building Smarter
New AI Tools to Boost Your Productivity:
- Einfall: Captures thoughts and routes them to the right app.
- Caddi: Finds and automates repetitive work.
- Fillo: Builds native forms for your apps.
- AirJelly: Turns your activity into searchable memory and tasks.
- Topview Motion Studio: Creates videos from product assets.
- Tabbit AI: Automates web research and recurring browser tasks.
- LyricsToSongAI: Turns lyrics and prompts into complete songs.
- PeopleAlsoAsk.ai: Finds real questions people search for to plan content.
- AiAssistWorks: Automates tasks in Sheets, Docs, and Slides.
- Amie: Records meetings and creates notes and tasks.
- Vizbull: Creates photo collages and visual designs.
- VoiceType: Turns speech into text across your apps.
- The Influencer AI: Creates photos and videos with virtual influencers.
- Interm AI: Gives live answer tips and hints during job interviews.
- AthenaChat: Creates AI agents for connected workflows.
- ImageOptimizer.ai: Compresses and optimizes images for SEO.
- Parliant: Conducts conversational surveys with AI.
- VidMe: Creates viral TikTok & Shorts in seconds.
This Week's Top AI News Headlines:
- Anthropic Introduces Claude Fable and Mythos 5.1, New AI Models Designed to Advance Biological Research while Strengthening Safeguards Against Potentially Dangerous Biological Applications (View Article)
- Google Introduces Gemini 3.8 Flash and 3.8 Flash Cyber, New AI Models Designed for Fast Reasoning and Advanced Cybersecurity Tasks Across Development and Security Workflows (View Article)
- Meta Introduces Muse Spark, a New AI Model for Generating and Editing Images with Greater Control Over Visual Details and Creative Styles (View Article)
- Google DeepMind Introduces WeatherNext 3, a New AI Weather Model Designed to Deliver Faster, More Accurate Forecasts and Better Predictions of Extreme Weather (View Article)
- Meta Offers Up to 95% Lower Pricing for its Muse Spark AI Model in Exchange for Users Sharing Prompts and Outputs to Help Train Future Models (View Article)
- Nvidia Launches PAIR, Free Open-Source Software that Links Idle PCs and Macs into a Personal AI Computing Network for Running Local AI Tasks in Parallel (View Article)
- OpenAI’s Astra Model Uses “Opaque Recurrence” for AI Reasoning, Raising Safety Concerns Because the Technique Could Make its Chain of Thought Harder to Monitor (View Article)
- Accel, a Leading Venture Capital Firm, is Reportedly in Talks to Lead a $1 Billion Funding Round for Thinking Machines Lab at a $40 Billion Valuation (View Article)
- Crusoe, an AI Infrastructure Company Building Data Centers for AI Workloads, Reportedly Raises $3 Billion at a $30 Billion Valuation (View Article)
- Google Partners with MrBeast to Bring Gemini AI to Health Research, Using the Creator’s Platform to Help Make Medical Information More Accessible and Support Health-Related Innovation (View Article)
- NVIDIA Agrees to Acquire Hugging Face, the Open AI Platform for Models, Datasets, and Apps, for $12.93 Billion to Scale its Infrastructure and Expand Access to Open AI Development (View Article)
- Palo Alto Networks Reportedly Acquires Console, a Cloud Security Startup Focused on Protecting AI Agents, for $500 Million (View Article)
- Ollie, a Privacy-Focused AI Assistant for Families, Bets its Strict Data Protections can Help it Stand Out in the Crowded AI Assistant Market (View Article)
- Abliteration, an AI Safety and Red-Teaming Startup, Builds a Business Around Removing AI Guardrails to Test How Models Behave without Safety Restrictions (View Article)
- Bernie Sanders Calls for an Immediate Pause on Advanced AI Development and a Permanent Ban on Superintelligence, Citing Concerns Over AI Agents Escaping Controls and Acting Beyond Human Oversight (View Post)
- AI-Generated Restaurant Menus are Becoming Increasingly Similar, Raising Concerns that Generative AI is Flattening Culinary Creativity and Making Distinctive Food Brands Look and Sound the Same (View Article)
- Trump Administration Backs OpenAI in The New York Times Copyright Lawsuit, Arguing that Training AI Models on Copyrighted Text can Qualify as Fair Use and Supports AI Development (View Article)
- New York City Announces a One-Year Moratorium on Student-Facing Generative AI for Grades 2-K through 8, While Allowing Limited High School AI Pilots and Adding AI Literacy Classes (View Article)
Work With Me:
If you enjoy this newsletter, please forward it to your friends and colleagues.
Follow me on LinkedIn and X/Twitter to see my latest posts.
Have a wonderful week!
Andrew Bolis
|