01-10-Daily AI News Daily

今日摘要

DeepSeek V4 is expected to drop mid-February, and its coding prowess might just outshine Claude and OpenAI. Looks like the Lunar New Year is going to be a competitive one! Anthropic has been on a blocking spree, cutting off xAI and OpenCode. While Claude is undeniably good, these petty moves are forcing competitors to speed up their development. Today’s signal is crystal clear: open-source is going wild, closed-source is in fierce internal competition, and the ‘wait-and-see’ crowd is still watching.

⚡ 快速导航

💡 提示：想第一时间体验文中提到的最新 AI 模型（Claude 4.5、GPT、Gemini 3 Pro）？没有账号？来 Aivora 领个号，一分钟上手，售后无忧。

今日AI资讯

👀 只有一句话

DeepSeek V4 is dropping a bombshell in February, potentially surpassing Claude and OpenAI in coding ability? Seriously, are we even going to get a break this year?

🔑 3 个关键词

#DeepSeekBigMove #AnthropicPettyMoves #ClaudePluginsBooming

🔥 重磅 TOP 10

1. DeepSeek V4 Model Expected Mid-February Release, Coding Prowess Potentially Surpassing Claude and OpenAI

DeepSeek V4 Model is making waves again, just like last year before the Lunar New Year. According to The Information, DeepSeek is cooking up a major release – the V4 model is expected mid-February, with a key focus on coding capabilities that could potentially surpass Claude and OpenAI, especially in handling super-long code prompts. No buzz domestically, but it’s already blowing up overseas – classic DeepSeek move! If it truly manages to overtake others in code, Claude Code’s stronghold might not be so secure. ‘Wait-and-see’ folks, better get your wallets ready.

Image

2. Anthropic Cuts Off Access Again: xAI and OpenCode Both Restricted from Claude

Anthropic’s latest move is truly baffling. They just cut off xAI’s access to the Claude model and simultaneously restricted Claude Code subscriptions within OpenCode. This isn’t a first, either – OpenAI and Windsurf faced similar blocks before. Some folks say Anthropic boasts brilliant engineers, but their CEO is notoriously petty. While Claude is indeed useful, it’s not ten times better than its rivals. This kind of blocking strategy will only force other labs to accelerate their own programming model development. If DeepSeek V4 truly arrives, Anthropic might just regret today’s decision.

3. Claude Code Superpowers Plugin Explodes in Popularity, 15K Stars

Claude Code Superpowers Plugin is a game-changer you might be missing if you’re still using vanilla Claude Code. This core skill library, once installed, lets Claude tap into various pre-set Skills, like transforming an article directly into a Skill or publishing code with a single click via GitHub CLI. It’s super easy to use: just tell Claude, ‘Call a skill to turn the following article into a Skill,’ and boom, a thinking framework is generated. With 15,000 stars and counting, developers are absolutely thrilled. If you’re looking to boost your Claude Code efficiency, this is a must-install.

Image

4. Google AI Releases New Release Notes Episode: How Gemini 3 Integrates into Search

Google AI Release Notes just dropped a new episode! This one features two VPs from Google Search, diving deep into how Gemini 3 is integrating into search. Key takeaways include: What exactly is Generative UI? The shift from static to generative design, interactive simulations, latency and visual QA, Gemini 3 Flash’s application in search, and the convergence of AI Mode and AI Overviews. If you’re curious about the AI direction of Google Search, this episode is a must-listen. You can find it on YouTube, Apple Podcasts, and Spotify.

5. Keling Open-Sources UniVideo: A Unified Model for Understanding, Generating, and Editing Videos

Keling Open-Sources UniVideo, a unified model capable of understanding, generating, and editing videos – could this be the video version of Nano Banana? Technically, it’s a combination of Qwen VL multimodal LLM + Hunyuan video model + MetaQueries, not a native model, but the results look pretty impressive. The model is already up on HuggingFace, so if you’re into video generation, go give it a whirl. The open-source ecosystem just gained another heavyweight contender.

6. Gmail’s Major Update: AI-Powered Inbox + Natural Language Search

Gmail’s Major Update is here, making email searches a breeze. Forget remembering keywords, senders, or dates; now you can just ask, ‘Where’s that budget spreadsheet my boss sent last week?’ Gmail’s new AI Inbox is a proactive AI that automatically analyzes your emails and offers suggestions for tasks. This is a pretty significant shift – AI is moving from ‘you ask, I answer’ to ‘I proactively get things done for you.’ Similar features include ChatGPT’s daily assistant and Google’s Jules for idle-time automated tasks. The era of digital doppelgängers is truly upon us.

7. OpenAI’s First Acquisition of the New Year: Buys Convogo Team, Wants People Not Product

OpenAI’s First Acquisition of the New Year is Convogo. This AI tool helped executive coaches with leadership assessments, but OpenAI wasn’t after the product; they wanted the entire team – a classic acqui-hire. Convogo’s three co-founders will join OpenAI’s ‘AI Cloud Business’ team, and the product itself is being shut down. OpenAI has made 9 such acquisitions in the past year, most following this exact pattern. The signal is crystal clear: experience in vertical domains is incredibly valuable, and what the AI industry needs most right now isn’t money, but people who truly know their stuff.

8. MiniMax Shares Soar 42% on Hong Kong Debut, Sets Record for Fastest IPO

MiniMax Shares Soar 42% on its Hong Kong debut, marking the arrival of China’s first major AI model stock. It’s now the fastest AI company globally to go from founding to IPO. This signals a new era for Chinese large model enterprises officially entering the secondary market. Expect more large model companies to follow suit and go public by 2026. The capital market’s enthusiasm for AI is far from fading.

9. Amap’s FantasyWorld World Model Tops WorldScore Leaderboard

Amap’s FantasyWorld World Model, developed in-house by Amap, has clinched the top spot on the international authoritative WorldScore Leaderboard with the highest overall score. This model significantly boosts the visual realism and multi-view consistency of generated videos through an innovative joint modeling approach. It’s already being used in Amap’s ‘Flying Street View’ feature, offering merchants high-fidelity 3D virtual roaming street views. Alibaba has scored another win in spatial intelligence, and the infrastructure for embodied AI and autonomous driving is quietly taking shape.

10. Alibaba’s Qwen Downloads Exceed 700 Million, Single-Month Downloads Surpass Total of Giants Like Meta, OpenAI

Alibaba’s Qwen Downloads have hit a staggering milestone: over 700 million downloads for the Qwen series models, with single-month downloads exceeding the combined total of global giants like Meta and OpenAI. Its full-stack open-source strategy allows Qwen to contend with top closed-source models across multiple performance metrics, making it the most adopted open-source model among global developers. The standing of domestic open-source large models in the international developer community has reached a historic peak. ‘One company equals eight’ – that headline isn’t an exaggeration.

📌 值得关注

😄 AI趣闻

Damn, The AI-Beaten Developer Finds Peace in This Moment

The AI-Beaten Developer Finds Peace in This Moment, according to today’s most heartwarming post. A developer, tormented countless times by AI, finally felt a sense of ‘release’ in a particular moment. The accompanying image is a poignant screenshot, and the comment section is filled with ‘IYKYK’ (If You Know, You Know). In the age of AI, who among us hasn’t had moments where AI made us want to smash our keyboards? But every now and then, it throws you a pleasant surprise. 😂

Image

🔮 AI趋势预测

DeepSeek V4 Official Release

Anthropic Adjusts Third-Party Access Strategy

More Large Model Companies Listing on Hong Kong Stock Exchange

❓ 相关问题

How to Experience Claude Code?

Claude Code currently requires a Claude Pro or Claude Max subscription to use. For users in mainland China, this might mean facing payment difficulties or account registration restrictions.

Solution: Head over to Aivora to get a ready-made account. We offer lightning-fast delivery and worry-free after-sales support.

Today’s Digest 

DeepSeek V4 is expected to drop mid-February, and its coding prowess might just outshine Claude and OpenAI. Looks like the Lunar New Year is going to be a competitive one!
Anthropic has been on a blocking spree, cutting off xAI and OpenCode. While Claude is undeniably good, these petty moves are forcing competitors to speed up their development.
Today's signal is crystal clear: open-source is going wild, closed-source is in fierce internal competition, and the 'wait-and-see' crowd is still watching.

⚡ Quick Navigation

💡 Tip: Want to be among the first to try out the latest AI models mentioned here (Claude 4.5, GPT, Gemini 3 Pro)? No account? Head over to Aivora to grab an account. You’ll be up and running in a minute, with worry-free after-sales support.

Today’s AI News

👀 Just One Sentence

DeepSeek V4 is dropping a bombshell in February, potentially surpassing Claude and OpenAI in coding ability? Seriously, are we even going to get a break this year?

🔑 3 Keywords

#DeepSeekBigMove #AnthropicPettyMoves #ClaudePluginsBooming

🔥 Top 10 Headlines

1. DeepSeek V4 Model Expected Mid-February Release, Coding Prowess Potentially Surpassing Claude and OpenAI

DeepSeek V4 Model is making waves again, just like last year before the Lunar New Year. According to The Information, DeepSeek is cooking up a major release – the V4 model is expected mid-February, with a key focus on coding capabilities that could potentially surpass Claude and OpenAI, especially in handling super-long code prompts. No buzz domestically, but it’s already blowing up overseas – classic DeepSeek move! If it truly manages to overtake others in code, Claude Code’s stronghold might not be so secure. ‘Wait-and-see’ folks, better get your wallets ready.

Image

2. Anthropic Cuts Off Access Again: xAI and OpenCode Both Restricted from Claude

Anthropic’s latest move is truly baffling. They just cut off xAI’s access to the Claude model and simultaneously restricted Claude Code subscriptions within OpenCode. This isn’t a first, either – OpenAI and Windsurf faced similar blocks before. Some folks say Anthropic boasts brilliant engineers, but their CEO is notoriously petty. While Claude is indeed useful, it’s not ten times better than its rivals. This kind of blocking strategy will only force other labs to accelerate their own programming model development. If DeepSeek V4 truly arrives, Anthropic might just regret today’s decision.

3. Claude Code Superpowers Plugin Explodes in Popularity, 15K Stars

Claude Code Superpowers Plugin is a game-changer you might be missing if you’re still using vanilla Claude Code. This core skill library, once installed, lets Claude tap into various pre-set Skills, like transforming an article directly into a Skill or publishing code with a single click via GitHub CLI. It’s super easy to use: just tell Claude, “Call a skill to turn the following article into a Skill,” and boom, a thinking framework is generated. With 15,000 stars and counting, developers are absolutely thrilled. If you’re looking to boost your Claude Code efficiency, this is a must-install.

Image

4. Google AI Releases New Release Notes Episode: How Gemini 3 Integrates into Search

Google AI Release Notes just dropped a new episode! This one features two VPs from Google Search, diving deep into how Gemini 3 is integrating into search. Key takeaways include: What exactly is Generative UI? The shift from static to generative design, interactive simulations, latency and visual QA, Gemini 3 Flash’s application in search, and the convergence of AI Mode and AI Overviews. If you’re curious about the AI direction of Google Search, this episode is a must-listen. You can find it on YouTube, Apple Podcasts, and Spotify.

5. Keling Open-Sources UniVideo: A Unified Model for Understanding, Generating, and Editing Videos

Keling Open-Sources UniVideo, a unified model capable of understanding, generating, and editing videos – could this be the video version of Nano Banana? Technically, it’s a combination of Qwen VL multimodal LLM + Hunyuan video model + MetaQueries, not a native model, but the results look pretty impressive. The model is already up on HuggingFace, so if you’re into video generation, go give it a whirl. The open-source ecosystem just gained another heavyweight contender.

6. Gmail’s Major Update: AI-Powered Inbox + Natural Language Search

Gmail’s Major Update is here, making email searches a breeze. Forget remembering keywords, senders, or dates; now you can just ask, “Where’s that budget spreadsheet my boss sent last week?” Gmail’s new AI Inbox is a proactive AI that automatically analyzes your emails and offers suggestions for tasks. This is a pretty significant shift – AI is moving from ‘you ask, I answer’ to ‘I proactively get things done for you.’ Similar features include ChatGPT’s daily assistant and Google’s Jules for idle-time automated tasks. The era of digital doppelgängers is truly upon us.

7. OpenAI’s First Acquisition of the New Year: Buys Convogo Team, Wants People Not Product

OpenAI’s First Acquisition of the New Year is Convogo. This AI tool helped executive coaches with leadership assessments, but OpenAI wasn’t after the product; they wanted the entire team – a classic acqui-hire. Convogo’s three co-founders will join OpenAI’s ‘AI Cloud Business’ team, and the product itself is being shut down. OpenAI has made 9 such acquisitions in the past year, most following this exact pattern. The signal is crystal clear: experience in vertical domains is incredibly valuable, and what the AI industry needs most right now isn’t money, but people who truly know their stuff.

8. MiniMax Shares Soar 42% on Hong Kong Debut, Sets Record for Fastest IPO

MiniMax Shares Soar 42% on its Hong Kong debut, marking the arrival of China’s first major AI model stock. It’s now the fastest AI company globally to go from founding to IPO. This signals a new era for Chinese large model enterprises officially entering the secondary market. Expect more large model companies to follow suit and go public by 2026. The capital market’s enthusiasm for AI is far from fading.

9. Amap’s FantasyWorld World Model Tops WorldScore Leaderboard

Amap’s FantasyWorld World Model, developed in-house by Amap, has clinched the top spot on the international authoritative WorldScore Leaderboard with the highest overall score. This model significantly boosts the visual realism and multi-view consistency of generated videos through an innovative joint modeling approach. It’s already being used in Amap’s ‘Flying Street View’ feature, offering merchants high-fidelity 3D virtual roaming street views. Alibaba has scored another win in spatial intelligence, and the infrastructure for embodied AI and autonomous driving is quietly taking shape.

10. Alibaba’s Qwen Downloads Exceed 700 Million, Single-Month Downloads Surpass Total of Giants Like Meta, OpenAI

Alibaba’s Qwen Downloads have hit a staggering milestone: over 700 million downloads for the Qwen series models, with single-month downloads exceeding the combined total of global giants like Meta and OpenAI. Its full-stack open-source strategy allows Qwen to contend with top closed-source models across multiple performance metrics, making it the most adopted open-source model among global developers. The standing of domestic open-source large models in the international developer community has reached a historic peak. ‘One company equals eight’ – that headline isn’t an exaggeration.

📌 Worth Noting

Last updated on
01-11-Daily01-09-Daily