Discover the AI Tools

Alibaba just dropped Qwen3.8-Omni-Flash, a native multimodal AI that handles text, images, audio, and video in one go. It boasts a million-token context window and slashes audio processing costs by over 98%. But can it really rival Google's Gemini? The benchmarks tell a nuanced story.

What happens when a trio of hackers armed with Claude and Codex subscriptions pokes at OpenAI's defenses? They slip into an employee's ChatGPT account, wander through private code, and walk away with a $6,500 bounty. Their takeaway: if three people can do this, imagine what a nation-state could do.

Zhipu AI just dropped GLM-5.3-FlashX, a model that spits out 200 tokens per second. It runs on a massive 100,000-unit domestic chip base, keeping costs low while boosting speed. Developers can tap into the API, and regular folks can test it in the experience center. This upgrade shows China's large models are getting faster and more accessible.

Andrew Ng isn't buying the AI apocalypse hype. In a recent interview, he pushed back against warnings from top AI labs, calling them more fiction than fact. Ng argues these narratives distract from real safety issues—like cybersecurity and misuse—and that we should focus on engineering solutions, not existential fear.

Anthropic just dropped a number that should make every AI lab sit up straight: Claude now handles 26% of its internal R&D. That's up from less than 1% in February. With 30,000 agents making a billion decisions a month, the race for recursive self-improvement is no longer theoretical. Competitors like OpenAI and Google are under pressure to show their own cards.

吴恩达最近忍不住开怼了：那些AI灭绝人类的警告，听起来更像科幻剧本，而不是科学结论。他直言，这波炒作背后可能藏着影响监管的小算盘。AI安全当然重要，但把精力全花在吓唬人上，还不如多解决点实际问题。

Tesla just dropped a software update that brings its Doubao voice assistant to more models, along with over a dozen upgrades. From pet mode to navigation, entertainment to safety, there's a lot to unpack. But some features require a premium subscription. Curious what's new? Let's dive in.

Chen Hang, the founder of DingTalk, has stepped down as CEO and taken a non-executive director role at Baiwang Co., Ltd. with an annual salary of just 100,000 RMB. This move comes three months after his departure from DingTalk, following a controversial resignation letter. What does this mean for the veteran entrepreneur and the tax AI company?

Zhipu just dropped GLM-5.3-FlashX, a high-speed AI model that cranks out up to 200 tokens per second. Built on the popular GLM-5.3-Flash, it's the first to nail intelligence, price, and speed all at once. Enterprise developers, get ready—this could change how you build.

Struggling to find the perfect spot for your store? Qwen Office has integrated Gaode's Store Operation Expert Kit, offering site selection consulting, mall insights, and store diagnosis. This AI-powered tool analyzes traffic, competition, and more, turning a week-long headache into a one-hour task. With over 30 million users, Qwen Office aims to give merchants a data-driven edge in physical retail.

Anthropic's own AI, Claude, now handles 26% of its R&D tasks—up from less than 1% in February. With 30,000 agents making a billion decisions a month, the company walks a tightrope between explosive automation and human oversight.

Calibre just got a playful update. Version 9.15 sneaks in an optional AI writing game called 'Create Your Own Adventure' — and it's off by default. You'll need to dig into settings to turn it on. Once activated, the AI builds a fictional world while you type to push the story forward. It's like reading a book, but you're the co-author. A clever twist for an e-book manager.

Crusoe just raised $3 billion, rocketing its valuation to $30 billion in under a year. The data center startup, which once mined crypto with flare gas, now powers AI for Meta and OpenAI. A new $1.3 billion cloud deal with Jane Street signals its growing clout. With IPO whispers swirling, Crusoe's story shows how AI's hunger for compute is minting billion-dollar businesses.

Google DeepMind just launched a new institute to dig into the hard problems of AGI—think economic policy, model transparency, and how we evaluate frontier AI. With big names like Demis Hassabis and Shane Legg on board, the group is already proposing bold ideas, like a US-led standards body and confidential testing. It's a clear sign that AGI conversations are moving from 'can we build it?' to 'how do we govern it?'

SpaceX's AI team is quietly exploring a new source of training data: the operational records of struggling or failed startups. The goal? Feed Grok with real-world business data beyond X and internal teams. It's a strategy Google has tried before, but can it work? No deals yet, and SpaceX isn't talking.

Google is secretly building Mathematica, a math-focused AI that could be its most powerful yet. Leaked API details reveal a 1 million token context window and a 65,536 token output limit, all tagged as an unstable experiment. But what does Google's internal 'Teamfood' testing mean for the future of AI reasoning?

Tencent quietly rolls out Chatterfly, a new AI voice assistant that turns your spoken words into polished text—no more rambling. Currently in internal testing for Mac and Windows, it promises local data storage and smart intent understanding. Could this be the end of typing as we know it?

Anthropic just gave Claude Code a big upgrade: multi-threaded parallel execution. Describe your project goal, and a coordinator splits the work across independent cloud sessions. You can track progress in real time, even on mobile. It's a step toward AI that truly collaborates with itself. Pro and Max users can test it now; team versions are coming.

OpenAI just rolled out Astra for Law, a specialized AI platform that taps into over 230 million legal URLs. It scores 54% on a tough legal Q&A benchmark—beating plain GPT-6Astra by 15 points. But don't expect it to argue in court. It's built to help lawyers dig up precedents, spot unfavorable rulings, and analyze contract risks. Think of it as a super-powered paralegal that never sleeps.

Ever dreamed of owning a humanoid robot? The Booster K1 Explorer Edition just hit JD.com for 38,999 yuan, and it's built for developers who want to tinker. Standing 95cm tall and weighing 19.5kg, it packs 22 degrees of freedom, dual-depth cameras, and a DouBao large model that lets it chat and understand you right out of the box. With 48TOPS of computing power and open APIs, this little bot could be your gateway to embodied intelligence.