China’s Moonshot and Alibaba dropped massive AI models, openly available to developers worldwide.
America keeps its best locked up. If openness wins, does secrecy become Silicon Valley’s weakness?
Today in the “Incredible AI” world:
Chinese Giants Pressure Silicon Valley Labs
Feyn AI Turns SQL Into Detective Work
Microsoft Bets Bigger On AMD AI Chips
How to Master Lead Attribution with AI
5 Super-Useful AI Tools You Should Try
Top AI-Generated Image of the Day
Read Time: 4 minutes
- LATEST DEVELOPMENTS -
China’s AI Models Challenge US Dominance:
Moonshot casually dropped a 2.8 trillion parameter model like it was a mixtape. Alibaba followed with 2.4 trillion, calling it “continuously evolving.” Both are open-source, which makes America’s closed-door strategy look oddly insecure.
Things to Know:
Industry Shockwave: Moonshot’s Kimi K3 claims to outperform nearly every US model except GPT-5.6 Sol and Claude Fable 5. Alibaba’s Qwen3.8 positions itself as second only to Anthropic’s flagship.
Open Source Gambit: Unlike US labs guarding their crown jewels, these Chinese models promise open weights. Developers worldwide can download, adapt, and remix frontier systems without Silicon Valley’s permission slip.
Geopolitical Undercurrent: Washington is tightening export controls, restricting chip access, and even forcing Anthropic to pull models. China’s openness flips the script, turning AI into a geopolitical chess match with unexpected moves.
Economic Reality Check: US firms burn billions on chips and training, betting size equals dominance. China’s rivals suggest scale isn’t everything, raising doubts about whether money alone secures lasting advantage.
This isn’t just about benchmarks or parameter counts. It’s about who controls the future of intelligence itself. If China’s open-source gamble sticks, the next wave of AI breakthroughs may not come from Silicon Valley at all.
This AI Actually Checks Its Homework:
A four billion parameter model just matched Claude Opus on a notoriously brutal SQL benchmark. Its secret isn't raw intelligence, it just asks first. Turns out humility scales better than raw brainpower.
Things to Know:
Checks First: SQRL inspects the database with read-only queries before committing to an answer. No more guessing ambiguous column names.
Numbers Don't Lie: SQRL's flagship model achieves 70.6 percent accuracy on BIRD Dev. Claude Opus 4.6 trails at 68.77 percent.
The Fun Part: The tiny four billion parameter version matches Opus outright. Your data never leaves your own servers.
Trails Behind: Sonnet, Qwen3-Coder, GLM, DeepSeek, and Kimi all scored lower. One specialized model just outclassed the entire field.
This isn't about SQL, it's about models finally admitting they don't know everything. A model that asks first beats one that guesses fast. That instinct might matter more than parameter count ever did.
The browser that reads the room before you ask.
Most browsers get you to the page. Norton Neo gets you to the answer. Magic Box understands your intent before you finish typing — no prompting, no switching apps, no copy-pasting. Built-in AI, instantly and for free. Privacy handled by Norton, by default.
AMD Launches Helios, Its First Rack AI System:
Nvidia controls 95 percent of the data-center GPU market, practically a monopoly. AMD just convinced Microsoft to place a serious wager. Sometimes the biggest underdog just needs one big name to say yes.
Things to Know:
Signed On: Microsoft will deploy Helios racks inside its Azure data centers. That makes four major buyers so far.
Math Is Lopsided: Nvidia holds more than 95 percent of the market today. AMD holds roughly 4.5 percent, for now.
Analyst Is Bullish: Daniel Newman thinks AMD could grab twenty to twenty five percent. That's hundreds of billions in revenue.
Real Money Ahead: AMD expects tens of billions in AI revenue by 2027. Most of that comes from Helios alone.
For years the entire AI chip conversation had exactly one company. That's no longer true, and hyperscalers finally forced the change. Monopolies rarely end with a bang, they usually end with a spreadsheet.
- TRY THIS -
How to Track and Qualify Every Lead with AI:
If you’re running marketing campaigns, you know the frustration — you’re getting calls, form fills, and chats, but you can’t clearly see which campaigns are driving real business.
WhatConverts is an AI-powered lead tracking and reporting platform that captures every lead, shows you where it came from, and helps you qualify it — all in one place.
How it works:
Add tracking code: Get the tracking code from your WhatConverts dashboard and add it to your website.
Capture everything: The platform automatically tracks all conversion types—phone calls, web forms, and live chats from every visitor.
Auto-organize leads: Every lead from every marketing campaign gets sorted and organized in one central dashboard where you can value and qualify them.
Get actionable insights: Smart reports show you exactly which marketing channels, keywords, and campaigns are delivering your best leads with complete attribution data.
Stop guessing which marketing campaigns are worth your budget. With WhatConverts, you can see exactly where your best leads come from and double down on what actually works.
- DAILY POLL AND RESULT -
Today’s Poll:
Q) Should AI tools own the data they improve from?
Vote and find out the result tomorrow.
Yesterday’s Result:
Q) Will AI agents replace traditional SaaS tools?
A) Yes, agents bundle workflows - 60% 👑
B) No, SaaS remains specialized - 40%
- TOOLS OF THE DAY -
Movio AI: Movio creates AI spokesperson videos with multilingual avatars and voiceovers for marketing and business communication.
Longform AI: Longform AI helps you research, outline, and write long-form articles and reports with a structured AI workflow from start to finish.
HuggingChat: HuggingChat is Hugging Face's free, open-source ChatGPT alternative powered by the best open-source language models available.
Podium Podcast: Podium generates podcast show notes, summaries, chapters, and social posts from your episode audio automatically using AI.
Kaedim: Kaedim uses AI to transform concept art or reference images into game-ready 3D models in just a few minutes.
- AI IMAGE OF THE DAY -
Snow Leopard Prowls The Neon Rain:
Prompt:
“Anthro snow leopard, fully fur-covered (no skin, furry hands and fingers), amber eyes, confident smirk, perched on a rain-soaked vending machine, katana with LED accents across one shoulder, flowing tail. Oversized cropped bomber jacket with dragon embroidery, straps, chains. Neon-lit cyberpunk alley, wet reflections, steam, mist. Semi-realistic painterly style, cinematic rim lighting, magazine cover quality.”
Try this prompt in any decent AI image generator and let me know your result.
