App也有聊天了

deepseek OCR仅仅是另一个OCR而已，并没有质的飞跃。还有那些人云亦云什么替代记忆机制，这种方式更像在闭着眼睛放枪，有效的概率不大，但试一试总比不试好。

Wait, what? This is much better than DeepSeek-OCR!

木牛流马都用上履带了

Gemini 3 Deep Think is here. Deep Think is our most advanced reasoning mode that explores multiple hypotheses simultaneously to give you an even more sophisticated output.

The US military successfully tested a high-power microwave weapon capable of neutralizing drones in seconds. In one trial, the system disabled 49 out of 61 target drones with a single deployment, using directed electromagnetic pulses to disrupt onboard electronics without

35 years later it still looks like the future.

如果美国趴下，世界将退回到中世纪，生产力更发达的中世纪。

Japan develops spherical solar cells that capture light from all directions. Japan has highlighted a solar technology called Sphelar, developed by Kyosemi Corporation, that uses tiny spherical silicon solar cells instead of flat panels. Each cell is about 1-2 millimeters wide

Everyone talking about mHC but no one yet talking about AltUp research.google/blog/alternati… (published 2023) and was part of gemma-3n.

Globe Observer The report is very misleading and even false in its dramatic form‼️ There is no evidence of secret Chinese military/cargo flights that would break through the restricted area without permission. Similar speculations about “mysterious Chinese cargo planes” appeared in June 2025

Amazing continual learning paper out of DeepMind 🚨 Most Continual Learning work assumes the backbone is fixed and the burden on the algorithm to fight catastrophic forgetting. This paper flips that assumption on its head and shows pretty convincingly that architecture choices

Robots don’t fail at the task. They fail at recovery. Cortex 1.5 introduces interactive policy patching: short human corrections + online RL → robots improve after deployment. No retraining. No rules. Just better behavior in production. 🔗 sereact.ai/posts/cortex-1…

有了Clawd → Moltbot → OpenClaw，是不是meta收购manus的几十亿白花了？

Claude Sonnet 5: The “Fennec” Leaks - Fennec Codename: Leaked internal codename for Claude Sonnet 5, reportedly one full generation ahead of Gemini’s “Snow Bunny.” - Imminent Release: A Vertex AI error log lists claude-sonnet-5@20260203, pointing to a February 3, 2026 release

感觉这张图可以值两白斤

The hallucination rate gap between proprietary and open-weight models remains very substantial on HalluHard (GPT-5.2-thinking = 53.8%, DeepSeek-Chat = 75.8%). Best open-weight models appear to be "benchmaxxed", and new benchmarks like ours reflect this. halluhard.com

How is this not big news?! Google DeepMind has unveiled Gemini Deep Think’s leap from Olympiad-level math to real-world scientific breakthroughs with their internal model "Aletheia", scoring up to 90% on IMO-ProofBench Advanced and tackling PhD-level problems with scalable

x.com/i/article/2023…

MIT-developed 3D printer can output a fully functional electric motor in a single process — team only needed to magnetize the linear motor after printing, motors cost just 50 cents each tomshardware.com/3d-printing/mi…