Models & Research

Anthropic interviewed 81,000 Claude users in the largest qualitative AI study ever conducted. The results are surprisingly honest.

Google DeepMind developed a framework measuring AGI progress through cognitive abilities. Now they are asking the community to help build the evaluations.

Google is rebuilding Maps around Gemini. The new 'Ask Maps' feature answers complex questions, the 3D navigation shows the world more realistically.

Anthropic merges three research teams into a new institute led by co-founder Jack Clark. The mission: figure out what AI does to society before it's too late.

Google brings Gemini in Chrome outside the US for the first time. India is the main focus with eight local languages, plus support for over 50 languages total.

Google brings Gemini deep into Docs, Sheets, Slides, and Drive. New features like 'Fill with Gemini' and AI drafts turn the suite into a true co-pilot.

Gemini 3 Pro Preview goes offline today. The short notice is frustrating developers.

Anthropic developed a new measure to track AI's real impact on the labor market. The findings are more nuanced than the headlines suggest.

Google launches the cheapest model in the Gemini 3 family. One-eighth the cost of Pro, configurable thinking depth, and faster than anything before it.

With the March 2026 Pixel Drop, Gemini can now operate apps on its own: ordering food, booking rides, reordering coffee. This isn't a chatbot anymore — it's an agent on your phone.

Alibaba releases four compact language models under Apache 2.0. From 0.8B to 9B parameters — and the 9B model beats OpenAI's open-source giant.

Google delivers five new features with the Gemini February Drop: music generation with Lyria 3, a significantly stronger Gemini 3.1 Pro, Deep Think for science, and more.

Google has expanded Gemini on Android with multi-step automation. The assistant can now independently book Ubers, order food, and compile shopping lists – initially on Pixel 10 and Galaxy S26.