01-09-Daily AI News Daily
Today’s AI News
Quick Navigation
- 📰 Today’s AI News – Get a quick overview of the latest happenings.
Pro Tip: Wanna be among the first to try out the latest AI models mentioned here (Claude 4.5, GPT, Gemini 3 Pro)? No account? No worries! Head over to Aivora to grab an account. You’ll be up and running in a minute, with hassle-free after-sales support.
Today’s AI News
👀 In a Nutshell
Google has made a decade-long comeback, Gmail has officially entered the Gemini era, and OpenAI is sounding the internal Code Red alarm.
🔑 3 Key Takeaways
#GoogleComeback #GmailTransformation #OpenSourceSurge
🔥 Top 10 Blockbusters
1. Wall Street Journal Long Read Reveals: How Google Overtook OpenAI
This ten-thousand-word article from the Wall Street Journal spills the beans on Google’s decade-long strategy. It all started at 2:30 AM in August 2025, when an image generator called “Nano Banana” quietly launched, skyrocketing to the top of the charts within days. By November, Gemini was crushing ChatGPT across multiple metrics, sending OpenAI into an internal “Code Red.” The article covers everything from Google’s self-developed TPU chips and DeepMind integration to the Bard flop and Sergey Brin’s return. The kicker? Google is reportedly negotiating with Meta to sell chips, a rumor that sent Nvidia’s stock tumbling 7% the day it broke.
2. Gmail Officially Enters the Gemini Era: Mailbox Transforms into an AI Assistant
Gmail has officially entered the Gemini era, transforming your mailbox into an AI assistant. Remember sifting through your inbox for ages to find important emails? Now, Gmail sorts them by priority for you. The AI inbox acts like a briefing page, pinning overdue bills and appointment reminders right at the top. The search bar now supports natural language queries, so you can ask things like, “Who was the plumber who quoted me for bathroom renovations last year?"—and Gemini will dig it up for you. Basic features are free, with advanced ones requiring a Pro/Ultra subscription. Currently, it’s rolling out in English in the US first. Your email is no longer just a tool for sending and receiving messages; it’s a whole new beast.
3. Alibaba Qwen Releases Multimodal Embedding and Reranking Models
Alibaba’s Qwen3-VL-Embedding and Qwen3-VL-Reranker models are here, unifying text, images, screenshots, videos, and mixed modalities into a single semantic space. What does this mean? Previously, these were all retrieved separately, but now you can search images and documents simultaneously with just one query. Multimodal RAG finally has a reliable open-source solution! These are the first batch of open-source models based on Qwen3-VL, specifically designed for cross-modal understanding.
4. Claude Code 2.1.0 Update: Skill Hot Reload is Here
Claude Code 2.1.0 just dropped an update, and Skill hot reload is finally here! What’s the most annoying thing when debugging skills? Having to restart after every single line of code change. Now, Claude Code supports hot updates, so your changes take effect immediately. You can also see your created skills with the “/” command, leading to more stable execution. No need to exit CC; just type claude update to get the latest. These are all pain points discovered during recent usage, and Anthropic’s iteration speed is seriously impressive.
5. Open-Source Search Agent MiroThinker 1.5: 30B Parameters Crushes Trillion-Parameter Models
MiroThinker 1.5, an open-source search agent, isn’t about massive parameter counts. Instead, it lets AI go wild researching, coding, and iterating like a human. This 30B parameter model has already surpassed ChatGPT-Agent in BrowseComp tests, all at just 1/30th of the cost! Its core tech, “Interactive Expansion,” means it doesn’t just guess when faced with uncertainty; it breaks down problems, retrieves evidence, and refines answers through multiple rounds. With over 2300+ Stars on GitHub, it’s a fresh deep research option for the open-source community.
6. Tencent Hunyuan Open-Sources HY-Motion1.0: 3D Character Animation Generation
Tencent Hunyuan’s HY-Motion1.0 is a godsend for game developers and animators! This billion-parameter model, built on a Diffusion Transformer architecture, can generate smooth, coherent skeletal animations. It covers over 200 action categories, including basic movements, daily life, and social interactions. Remember those late nights tweaking keyframes for 3D animation? Well, the barrier to entry just got significantly lower.
7. Zhipu AI Listed on Hong Kong Stock Exchange: Opens Up 3%, Market Cap Exceeds HK$43 Billion
Zhipu AI has officially listed on the Hong Kong Stock Exchange, opening up 3% with a market cap exceeding HK$43 billion! This marks a new phase for domestic large model companies, where commercialization is key. While R&D investment continues to climb, turning that into actual commercial success remains the critical challenge. The capital market has given a nod to its technical barriers and long-term potential, so now it’s all about whether they can make that money back.
8. Cerebras Lists GLM 4.7: One Word, Fast
Cerebras’s big advantage is its inference speed, and now GLM 4.7 is running on it too! This is awesome news for scenarios demanding low-latency responses.
9. Open-Source Android AI Chat Software TChat Updates to v1.7
TChat, the open-source Android AI chat client, just got an update to v1.7! It’s super clean and straightforward, letting you use your own API key. With support for various models, it’s perfect for developers who love tinkering with AI on their phones. It’s open-source on GitHub, and you can grab the release directly.
10. DIY Document API to Claude Code Skill: Easily Research Social Platforms Across the Web
You can now turn TikHub’s API documentation into a Claude Code Skill, letting you research data from Douyin, TikTok, Xiaohongshu, Instagram, YouTube, and more with just one sentence! Wanna know today’s hot topics on Xiaohongshu? Or search for DeepSeek discussions across the entire web? Just chat with it! The open-source address has been released, making it perfect for entrepreneurs looking to dive into social media research.
📌 Worth Your Attention
- [Product] OpenAI Releases ChatGPT Health - This health data management platform supports connecting to electronic medical records, with data not used for training.
- [Product] Alibaba Cloud Launches Multimodal Interaction Development Kit - Integrating Tongyi Qianwen, Wanxiang, and Bailing, it empowers AI glasses and robots.
- [Product] Yutori: Let AI Discover and Track Your Interests - It automatically generates monitoring workflows and is currently in beta testing.
- [Business] Douyin Establishes Second Headquarters in Shenzhen Nanshan - It will focus on pan-video R&D and AI labs.
- [Business] Jia Yueting Announces FF’s Entry into Embodied AI Robotics - The FX Super One will begin deliveries in Q2.
- [Open Source] NVIDIA alpasim Open-Sourced - A Python project with 517 Stars.
- [Open Source] VideoRAG: Chat with Videos - This is the accompanying code for a KDD'2026 paper, boasting over 2000+ Stars.
- [Research] BabyAGI Developer’s Ten-Thousand-Word Article: The State of AI in 2026 - It argues that value is shifting from interfaces to infrastructure, with governance as a prerequisite.
😄 AI Funnies
Xiamen Lab UNITY Model Logic Test Fails
The Xiamen Lab’s UNITY model, which claimed to surpass many domestic and international large models, totally fumbled a classic logic puzzle! The question was: “If you don’t put the red ball in the blue box, does that mean there’s definitely no red ball in the blue box?” The correct answer is “Not necessarily,” because “I didn’t put it in” doesn’t mean “it wasn’t there already” or “someone else didn’t put it in.” UNITY’s reasoning process looked super comprehensive and well-organized, but its final answer was “definitely not”—proving that no matter how long your chain of thought, a wrong answer is still a wrong answer. 🤦♀️
🔮 AI Trend Predictions
Gmail AI Features Go Global
- Predicted Timeline: Q1 2025
- Probability: 80%
- Basis: Today’s news about Gmail Officially Entering the Gemini Era + Google typically rolls out to other regions within 1-2 months after a US launch.
Multimodal RAG Applications Explode
- Predicted Timeline: Q1-Q2 2025
- Probability: 75%
- Basis: Today’s news about Qwen3-VL-Embedding Released + The maturity of open-source multimodal embedding models will accelerate enterprise adoption.
Claude Code Market Share Keeps Growing
- Predicted Timeline: Q1 2025
- Probability: 70%
- Basis: Today’s news about Claude Code 2.1.0 Update + Baoyu Complaining About Slow Codex CLI Iteration indicates Anthropic’s iteration speed is clearly leading the pack.
Google’s In-House Chip Business Goes Independent
- Predicted Timeline: Q2 2025
- Probability: 55%
- Basis: Today’s news about Google Negotiating with Meta to Sell Chips + Ironwood chips have already established a competitive edge.
❓ FAQs
How to Try Out Gemini 3 Pro?
Gemini 3 Pro is Google’s latest flagship model, totally outperforming ChatGPT on multiple metrics. Right now, you’ll need a Google AI Pro or Ultra plan subscription to get higher access. For users in mainland China, you might hit some snags with payments or account registration.
Solution:
- Aivora offers ready-made Gemini Pro accounts.
- Get instant delivery, use it right after ordering, and skip all the payment and registration headaches.
- They provide stable, dedicated accounts with worry-free after-sales support.
Visit aivora.cn to check out their full list of AI account services.
How to Try Out Claude Code?
Claude Code, Anthropic’s smart programming tool, now has version 2.1.0 with handy features like Skill hot reload. To get the full experience, you’ll need a Claude Pro subscription.
Solution:
- Aivora offers ready-made Claude Pro accounts.
- Get instant delivery, use it right after ordering, and skip all the payment and registration headaches.
- They provide stable, dedicated accounts with worry-free after-sales support.
Visit aivora.cn to check out their full list of AI account services.
Today’s Summary
Google has completed its decade-long comeback, with Gemini crushing ChatGPT on multiple metrics, and OpenAI sounding the internal ‘Code Red’ alarm. Gmail has transformed into an AI assistant, offering natural language email search and smart priority sorting. Your inbox is no longer just a send-and-receive tool. It’s been a day of epic AI showdowns! We highly recommend checking out the Wall Street Journal’s ten-thousand-word deep dive.
⚡ Quick Navigation
- 📰 Today’s AI News – Get a quick overview of the latest happenings.
Pro Tip: Wanna be among the first to try out the latest AI models mentioned here (Claude 4.5, GPT, Gemini 3 Pro)? No account? No worries! Head over to Aivora to grab an account. You’ll be up and running in a minute, with hassle-free after-sales support.
Today’s AI News
👀 In a Nutshell
Google has made a decade-long comeback, Gmail has officially entered the Gemini era, and OpenAI is sounding the internal Code Red alarm.
🔑 3 Key Takeaways
#GoogleComeback #GmailTransformation #OpenSourceSurge
🔥 Top 10 Blockbusters
1. Wall Street Journal Long Read Reveals: How Google Overtook OpenAI
This ten-thousand-word article from the Wall Street Journal spills the beans on Google’s decade-long strategy. It all started at 2:30 AM in August 2025, when an image generator called “Nano Banana” quietly launched, skyrocketing to the top of the charts within days. By November, Gemini was crushing ChatGPT across multiple metrics, sending OpenAI into an internal “Code Red.” The article covers everything from Google’s self-developed TPU chips and DeepMind integration to the Bard flop and Sergey Brin’s return. The kicker? Google is reportedly negotiating with Meta to sell chips, a rumor that sent Nvidia’s stock tumbling 7% the day it broke.
2. Gmail Officially Enters the Gemini Era: Mailbox Transforms into an AI Assistant
Gmail has officially entered the Gemini era, transforming your mailbox into an AI assistant. Remember sifting through your inbox for ages to find important emails? Now, Gmail sorts them by priority for you. The AI inbox acts like a briefing page