Google Chrome AI Browser

Typing is a thing of the past

Typeless turns your raw, unfiltered voice into beautifully polished writing - in real time.

It works like magic, feels like cheating, and allows your thoughts to flow more freely than ever before.

With Typeless, you become more creative. More inspired. And more in-tune with your own ideas.

Your voice is your strength. Typeless turns it into a superpower.

Download for Mac today

🌐 Meet Gemini in Chrome: Help right where you need it - LINK

Google added a new AI assistant called Gemini in Chrome that helps you while you browse. It can summarize videos, find your past tabs, and pull up info using what’s already open in your browser.

You can use something called AI Mode in Chrome to ask harder questions based on what you’re looking at, and it tries to give answers faster. It works with Google tools like Calendar, Lens, etc. 

They also made security and privacy a big deal— you control what Gemini in Chrome sees, you can pause it or delete what it has accessed. Google also gives protection tools like automated password changes when needed. 

⚡ xAI debuts a faster and more cost-effective version of Grok 4 - LINK

xAI released Grok 4 Fast, a version of Grok 4 that’s faster and much cheaper. It gives almost the same performance but uses 40% fewer tokens and cuts costs up to 98% on benchmark tasks.

It still handles reasoning well, meaning it can think through problems or understand tasks, while also giving quick answers for simpler stuff. It’s also available for free users on web, iOS, Android, plus via API.

Another big point: it has a 2 million token context window, which means it can look at and use a lot more past content (like long texts) without forgetting. 

🛠️ OpenAI introduce upgrades to Codex - LINK

OpenAI upgraded Codex with a new version called GPT‑5‑Codex. It’s faster, more reliable, and works better when collaborating in real time—whether you’re coding on a phone, web, or terminal.

It also adapts: some tasks need few seconds and simpler reasoning, while others may need hours of complex thinking. The model shifts its effort depending on task needs.

The upgrades help developers build things more smoothly, catch errors, and make Codex feel more like a coding partner than just an assistant. 

🤖 Alibaba introduce open-source deep research ai agent - LINK

Alibaba introduced Tongyi‑DeepResearch‑30B‑A3B, an open-source AI agent with 30 billion parameters, but only about 3 billion get activated per token. That keeps it lighter while still powerful.

It’s made for “long-horizon, deep information-seeking tasks,” meaning it’s good at digging through a lot of info over time and keeping context, not just quick replies. It performs well on many benchmarks used to test such abilities. 

Because it’s open-source, many developers can use it, inspect it, and maybe even improve it. This gives access and transparency that closed models often don’t. 

🎬 Luma AI introduced Ray3 - LINK

Luma AI introduced Ray3, a video model that can reason with visuals (not just generate images). It’s the first model to make 16‑bit HDR video outputs, which gives better color and quality. 

It has something called Draft Mode to let creators try ideas faster and make revisions easier. It’s made to help storytellers, studios, etc., create richer, more realistic video content. 

Ray3 is part of Luma’s push to build tools that work not only with text or images, but mix video, audio, visuals, motion—giving creators more expressive power.