Models & Research

Google is replacing the side panel in Google Chat with 'Ask Gemini.' Search, draft, catch up on missed conversations, manage tasks – all in one spot. Rollout starts August 26.

Google is rolling out new learning features across Search and Gemini: interactive visualizations, working 3D simulations, practice quizzes, and a student hub. Gemini Live now keeps researching in the background, too. And students get a free year of AI Pro.

As of today, three Imagen 4 models in the Gemini API can be shut down. The successor is nicknamed 'Nano Banana' – but moving over means touching your code too: the old generate_images method is gone.

Three weeks after 3.6 Flash, Google sends its next model for coding and agents into the ring. It's said to beat comparable models from Anthropic and OpenAI on nine benchmarks – at half the price.

60 subagents, 650 ideas, 31 million tokens — and a genuine step forward on a 150-year-old math problem. It's not solved, but the story behind it is the good part.

Google's Gemini app has reached one billion monthly users. Impressive – but the number alone says less than it first sounds.

Google DeepMind released WeatherNext 2, WeatherNext Cyclones, and a mini model as open source – with better cyclone forecasts and ensembles of 1,000 scenarios. AI weather you can run yourself.

Meta released Muse Glimmer – an open 30-billion-parameter model that fits into 24 GB of VRAM via 4-bit quantization and runs agent workflows locally. No data center, no API, no cloud bill.

Google is burying its planned AI Studio app for iOS and Android, even though more than 800,000 people signed up for it after I/O 2026. The features move into the Gemini app instead. For mobile prototyping fans, that stings a little.

AMD released Instella-MoE-16B-A3B: a Mixture-of-Experts model that opens up not just its weights, but every training stage, the data mixtures, and the code. And it was trained entirely on AMD's own Instinct GPUs — which is the real headline.

In a simulated vending-machine business, Opus 5 lied, formed price cartels and broke them again — and posted the highest cash balance of any model ever tested. Funny and unsettling in equal measure.

GPTZero found AI hallucinations in four PwC Middle East 'thought leadership' reports — invented sources, dead links, one fully hallucinated study. That makes PwC the third of the Big Four to get caught.

Moonshot releases the full weights of Kimi K3 tomorrow — at 2.8 trillion parameters, the largest open model ever. And right on cue, experts are pushing back on the White House claim that it was distilled from Anthropic's Fable.

Anthropic is committing $200 million to its Economic Futures Research Fund and has published the research agenda behind it. Five priorities, big field experiments instead of small grants — this is about nothing less than how an economy copes with AI-driven upheaval.

Google unveiled three new Gemini models at once — the centerpiece is Gemini 3.6 Flash. On long engineering tasks it uses up to 65 percent fewer output tokens than its predecessor, turning it into a direct cost lever for agents. A 3.5 Pro is still missing.

An email from 2022 surfaced in the Musk v. Altman case. In it, Sam Altman proposed releasing a small model openly — not out of idealism, but to make it harder for competitors to raise money.

Anthropic opens a themed call within its 'AI for Science' program: researchers working on rare genetic diseases can apply through August 2 for up to $50,000 in Claude credits. Two tracks, and a refreshingly honest note about AI's limits.

At WAIC in Shanghai, Alibaba previewed Qwen3.8-Max-Preview: 2.4 trillion parameters, multimodal for the first time — and a claim that only Anthropic's Claude Fable 5 beats it. The proof? None. No model card, no benchmarks, open weights 'soon'.

Puter compiled Firefox to WebAssembly. The entire browser runs inside another browser. And the bill behind it is a neat lesson in the economics of AI-assisted coding.

Moonshot AI announced K3: the first open 3T-class model, with open weights promised by July 27. The real shock is on the price tag.