There you have it! Alibaba just dropped Qwen3.8-Max, and it's already the #2 model on the planet …
August 4, 2026 · 0 likes · 0 comments
China Threat Workforce AI
There you have it! Alibaba just dropped Qwen3.8-Max, and it's already the #2 model on the planet for visual reasoning — behind only Anthropic. A Chinese open-weight model. Not "catching up eventually." Now.
UnbiasedHeadlines.com has the numbers: https://lnkd.in/efPQ4Ffu
2.4 trillion parameters. Ran a software project by itself for 16 days straight in testing. Weights go public for free download next week.
And while Alibaba chases the benchmark crown, DeepSeek is knocking the floor out from under everyone on price. Their new V4-Flash costs about 3 cents to run a full test suite. Anthropic's top model? $3.15. That's more than 100 times cheaper on the same work.
Read the strategy underneath that. One Chinese lab races you on capability. The other races you to zero on cost. And they're doing it on a fraction of the chips we spent years trying to keep out of their hands.
I wrote a whole chapter of REPLACEMENT on exactly this. It's titled "China Doesn't Debate. China Deploys." The line in it is simple: They deployed. We debated. China deploys. America debates. The clock doesn't care which one we choose.
That was supposed to be a warning. It's now Monday's headline.
On a lighter note — thank you. REPLACEMENT just hit #1 Amazon bestseller in TEN categories: Artificial Intelligence, Computers & Technology, Generative AI, Expert Systems, Information Technology, Business Technology Innovation, Business Office Skills, Computer Science, Office Automation Skills, and Business Education & Reference. That's on you. I'm genuinely floored.
If the book meant anything to you, do me one favor: leave an honest Amazon review. Reviews are what put a book in front of the people who still think this is years away. It isn't.
REPLACEMENT — available now in every format:
Ebook — https://lnkd.in/gZCFR5Tk
Still think we're winning this?
UnbiasedHeadlines.com has the numbers: https://lnkd.in/efPQ4Ffu
2.4 trillion parameters. Ran a software project by itself for 16 days straight in testing. Weights go public for free download next week.
And while Alibaba chases the benchmark crown, DeepSeek is knocking the floor out from under everyone on price. Their new V4-Flash costs about 3 cents to run a full test suite. Anthropic's top model? $3.15. That's more than 100 times cheaper on the same work.
Read the strategy underneath that. One Chinese lab races you on capability. The other races you to zero on cost. And they're doing it on a fraction of the chips we spent years trying to keep out of their hands.
I wrote a whole chapter of REPLACEMENT on exactly this. It's titled "China Doesn't Debate. China Deploys." The line in it is simple: They deployed. We debated. China deploys. America debates. The clock doesn't care which one we choose.
That was supposed to be a warning. It's now Monday's headline.
On a lighter note — thank you. REPLACEMENT just hit #1 Amazon bestseller in TEN categories: Artificial Intelligence, Computers & Technology, Generative AI, Expert Systems, Information Technology, Business Technology Innovation, Business Office Skills, Computer Science, Office Automation Skills, and Business Education & Reference. That's on you. I'm genuinely floored.
If the book meant anything to you, do me one favor: leave an honest Amazon review. Reviews are what put a book in front of the people who still think this is years away. It isn't.
REPLACEMENT — available now in every format:
Ebook — https://lnkd.in/gZCFR5Tk
Still think we're winning this?