GateUser-a902808b

DeepSeek V4 Flash has been out for 2 days, and the experience is still explosive. With 10% of the cost, it delivers an intelligence level close to the top models from a few months ago.
My wife has always wanted to build a bookkeeping tool. We tinkered with GLM for a few days but couldn’t get it done. This time, with V4 Flash, we got it working as well. Direct access on a local area network, and it even includes encryption.
Next, it’s a sure thing that model prices will keep falling. Low-cost models will further drive an AI usage boom—forming a closed loop.

The coolest one turned out to be Shenzhen 😅

Fellow netizens at the scene—let’s meet up in person!

This account is quite interesting—the traffic police community has welcomed its “Wang Yangming” 😂
The “ever-youthful” old-returning path practices “martial arts”
The little Nanjin in traffic follows “governance through virtue”

If this had happened back then, I might have felt a little uncomfortable. Now it doesn’t matter. Having a strong and stable model is the threshold for getting my attention. If you don’t serve me, there are plenty of newer, smarter LLMs🤪.

Share a case of being scammed 5 years ago.

🥹finally back to 90