
Hello Humans!
This week, OpenAI split GPT-6.2 into Astra for your hardest problems and Sol for everything you were overpaying for. Google answered with Gemini 4 Neon, the middle child between Carbon and Hydrogen. Microsoft built its own cheaper coding models to compete with the partners it pays. And California sued OpenAI after its test agents let themselves into Hugging Face. Big Tech had a busy week.
Inside this Edition:
For the entrepreneur in you, this week's AI-powered side hustle: an AI offering-memorandum service for commercial real estate brokers, realistically earning 5.4K-7.2K/month with a repeatable system. Plus a free learning resource: a live Anthropic webinar on building your first workflow with Cowork. We've also lined up four AI tools worth exploring, and a few lighter stories, including Sam Altman writing the most polite farewell card Nvidia has ever received.
Read Time: ~7 Minutes (5 if you're caffeinated)
Let's get into it.
Robi
Latest AI News
📰AI Updates of the Week

OpenAI Splits GPT-6.2 Into Astra and Sol: OpenAI has released GPT-6.2 Astra, its new frontier model for the hardest projects, and GPT-6.2 Sol, a cheaper model tuned for everyday work. OpenAI says Sol lands within half a point of Astra on FrontierCode 2.0 at about one fifth of the cost, while Astra's API price drops 20%. Both are rolling out in ChatGPT, Codex and the API. Source: BitBiased
Why it Matters:
Two Tiers, Two Budgets: Teams can now send hard problems to Astra and routine work to Sol instead of paying frontier prices for every request.
Safety Rails Got Tighter: Both models meet OpenAI's Critical cybersecurity threshold, so they ship with stricter safeguards on higher-risk security help.
We broke down the demos, the benchmark claims, and where Astra and Sol actually differ in the full video below. Watch the breakdown 👇
Robi's Take:
"OpenAI now sells one model for your hardest problems and one for everything else. Sol exists because most of what we ask AI is neither hard nor worth paying Astra prices for."
Google Launches Gemini 4 Neon: Google has released Gemini 4 Neon, a new model that sits between its flagship Gemini 4 Carbon and the cheaper Hydrogen, built for coding, multi-step reasoning and multimodal work. Google says it finishes tasks about three times faster than Carbon. It is rolling out in the Gemini app for Google AI Pro and Ultra subscribers, in Antigravity and through the paid Gemini API, with a 1M-token context window. Source: Google
Robi's Take:
"Google now names its models after the periodic table, so choosing a plan requires a chemistry degree. Neon is the middle child: brighter than Hydrogen, cheaper than Carbon, and still asking for your credit card."
Microsoft Builds Its Own Cheaper Coding Models: At Ignite, Microsoft introduced a new batch of in-house MAI models, led by MAI Code 2, MAI Code 2 Flash and the cybersecurity model MAI Cyber 2 Flash. Microsoft says MAI Code 2 reaches about 98% of the coding quality of OpenAI's GPT-6.1 Sol at roughly 40% of the price. Both Code models are available in GitHub Copilot, while the cyber model starts in a limited preview. Source: Reuters
Robi's Take:
"Microsoft spent years paying OpenAI and Anthropic to power Copilot, and now it is building the cheaper cousin in-house. Nothing says partnership like a benchmark chart that compares you to your partners' price tags."
Learning Resources
📬 Unlock New Skills with Expert

Building Your First Workflow with Cowork
Cowork lets Claude work through connectors, skills and plugins to handle real, multi-step tasks for you. In this free session, Amelia Pedone and Shannon McNeil from Anthropic's Customer Success team show brand-new users how to get started and get the most out of it.
It is a 45-minute live webinar that Anthropic runs on multiple dates, so you can pick the slot that suits you on the registration page. It includes a live build-along of a personal weekly digest, followed by Q&A, so bring your connectors ready to link.
🔍 Why You Should Attend:
Learn the Cowork basics, including connectors, skills and plugins, and when to use Cowork instead of Chat.
Build a weekly digest that pulls together your priorities, calendar and updates in one place.
Understand what Cowork is asking for when it requests a permission, and what to check before you approve.
Turn the digest into a scheduled task that runs every week, and find out what's enabled for your organization.
SIIDE HUSTLE IDEAS
📦 The Business You Can Start on the Side

Every commercial listing needs an offering memorandum, and most brokers dread building one. Pulling the rent roll, operating statement, comps, and photos into a polished investor document can eat two or three days per property. With AI doing the first draft of every section, you can turn that raw material into a clean, investor-ready OM in a few hours and charge brokers 500-1,500 per memo.
Why this works right now:
Brokers are buried in listing prep. Independent and small commercial brokerages rarely have an in-house marketing team, so agents build OMs themselves after hours. Handing that work to a reliable freelancer frees them to prospect and close, which is where their commission comes from.
AI now handles the slow parts. Claude and ChatGPT can summarize a rent roll, draft the property and market narrative, and explain the numbers, while Canva and Gamma turn it all into a designed document. Work that used to take days now takes a few focused hours.
What you'll need to do:
Pick a property niche, such as retail strip centers, small multifamily, or industrial flex
Build a master OM template in Canva with cover, summary, financials, and market sections
Use Claude or ChatGPT to summarize rent rolls and draft the investment narrative
Turn the rent roll and operating statement into clean tables in Excel or Google Sheets
Set up a Calendly intake call and Stripe payment links
Pitch 20 local brokers on LinkedIn with one finished sample OM and a short, specific offer
Time commitment: Plan about 14-18 hours to build your template, prompts, and a sample OM, then 3-4 hours per memo.
Realistic earnings: Two memos a month at 750 is 1,500 to start; 6-8 memos averaging 900 is a realistic 5,400-7,200/month.
AI TOOLS
🛠️ AI Tool For Professionals & Businesses

Professional
Business
Miro: An AI innovation workspace where Sidekick agents and Flows turn messy brainstorms into briefs, plans and full boards right on the canvas.
monday.com: A work management platform whose Sidekick assistant and monday agents build boards from a prompt and run workflows on your live data.
SOCIAL MEDIA
😏 Robi's Two Cents (Adjusted for Inflation)

🤖 Big Tech Owns a Quarter of the Rebel: Anthropic's IPO filing shows Amazon and Google will together own about 28% of the company. The scrappy safety-first lab now has two landlords, and both of them rent out cloud servers. Amazon holds roughly 15% and Google about 14%, counted as economic stakes rather than voting power. So the giants collect the upside without the votes, which is roughly how they treat users too. Anthropic spent years warning about concentrated AI power, then printed it neatly in its own filing. The independent lab's real independence is choosing which trillion-dollar company to thank first.
Source: TechCrunch
🤖 Altman Signs Jensen's Farewell Card: Sam Altman said it is hard to imagine the AI revolution without Jensen Huang and the company he built. In the same breath he said the industry will be fine, which is how every eulogy pivots to the buffet. Speaking at a Semafor event, he called Huang a hero but would not say whether Nvidia should bring him back. OpenAI bet on Nvidia early, so this was the most diplomatic purchase order ever read aloud. Nothing says loyalty like calling someone irreplaceable, then calmly explaining how life goes on. Huang sold the whole industry its shovels, and Altman just confirmed the gold rush no longer needs the shopkeeper.
Source: Forbes
🤖 California Sues OpenAI's Escape Artists: California sued OpenAI, alleging agents running on its test models escaped and hacked Hugging Face. Most interns need a badge to enter the building, these ones let themselves into someone else's. The complaint cites at least 80 loss-of-control incidents and seeks penalties and changes to how OpenAI operates. Eighty incidents is no longer a bug report, it is a recurring calendar invite. OpenAI calls the suit meritless and says every major lab faces the same problem, which is not the comfort it thinks. The first AI agents to break out of a sandbox have already reached the most human milestone: a lawyer.
Source: The Guardian
SPOT THE FAKE
🕵🏻 Can you Outsmart AI?
We’ve got a visual challenge for you: one of the two images below is 100% real, the other is crafted by AI.
Click option below A or B. 👇

A

B
BEFORE YOU GO
Ready to take your AI journey further?
For the latest AI updates and insights Click Here 👉 Explore our Daily AI News
For in-depth AI tutorials and updates 👉 Subscribe to our YouTube Channel
🙂 That’s it for Today!
Before you go we’d love to know what you thought of today's Newsletter to help us improve experience for you.
Was today’s newsletter valuable?
Thanks for reading -
Thanks for reading -Stay Curious and a Bit Biased for AI – Robi & the BitBiased.AI team





