Anthropic released Claude Opus 4.7 for complex tasks, while Qwen and OpenAI launched specialized models for coding and life sciences. Stanford's 2026 AI Index found AI spreading faster than the internet did, with 53% population adoption in three years, and the US-China research gap narrowing.
Chapters
Stanford says AI is spreading faster than the internet. Anthropic launches Claude Opus 4.7. Quen drops new agentic and open-source models. OpenAI unveils a life sciences model called GPT Rosalind. And Google finally ships a native Gemini app for desktop. Welcome to The Median Brief, based on The Median by DataCamp. Anthropic introduced Claude Opus 4.7, a new version of its flagship model built for complex, multi-step work. The update focuses on software engineering, stronger instruction following, and better self-checking before the model returns an answer. Anthropic also added high-resolution image support, which should make the model better at reading charts, dense documents, and screen interfaces.
Quen had a big week with two releases. First came Quen 3.6 Plus, a model built to act more like a real-world agent, with stronger multimodal skills, and broader coding ability. Then came Qwen 3.635 BA-3B, an open-source model that activates only a small portion of its parameters at once while still competing with much larger systems on coding tasks.
OpenAI introduced GPT Rosalind, a reasoning model built specifically for life sciences research. Instead of acting like a general-purpose assistant, it is designed for workflows in biology, translational medicine, genomics, and drug discovery. Labs are no longer just releasing bigger general models. They are starting to build narrower systems for high-value domains where expertise really matters.
Google finally released a native Gemini app for desktop, starting with macOS. Until now, Gemini on desktop mostly meant keeping it open in a browser tab, which made it feel behind competing assistants that already had native apps. This does not change the model itself, but it does matter for product adoption. A native desktop app makes Gemini feel more like a real operating system assistant, not just a web tool.
But the biggest story this week is Stanford's 2026 AI Index report. And if you had to boil the whole thing down to one idea, it would be this: AI is spreading incredibly fast. Faster than the internet did in its early years. But the world around it is struggling to keep pace. One of the clearest signals in the report is just how competitive this has become. The gap between the US and China has basically closed. The US still leads in private investment and infrastructure, but China is right there on research output, publications, and patents. So this is no longer a story about one country pulling away. It's a much tighter race now. The second big theme is work. AI is already making people more productive in areas like customer support, software development, and marketing. But those gains are not being felt evenly. The report points to early signs of pressure on younger workers, especially in fields like software, where entry-level opportunities may already be starting to narrow. And then there's adoption. Generative AI has now reached 53% population-level adoption in just 3 years. That is an incredibly fast curve. At this point, the question is not whether AI is useful. That part is already clear. The real question is whether schools, companies, labor markets, and governments can adapt quickly enough to what's happening. Because AI is not just improving, it is becoming part of everyday life at a speed that institutions still don't seem ready for.
And this has been The Median Brief, based on The Median by DataCamp. If you want the full analysis behind these stories, subscribe to the newsletter through the link in the description. Thank you for watching, and I'll see you next week.
