Sent to 20 readers
Race to AGI Daily Digest - Thursday, June 11, 2026
Share:
TLDR
Google’s DiffusionGemma shows how much headroom remains in making open generative models faster and cheaper to run.
Meituan’s Tabbit AI browser turns the desktop itself into an agent host, echoing the global shift toward always‑on assistants.
Anthropic’s $200M fund and Poetic’s $50M round show serious money flowing into understanding and managing AI‑driven job automation.
The Full Story
Following Monday’s massive Google–SpaceX–xAI GPU deal and the agentic assistant pushes from Apple and Anthropic, today feels like the next layer of the stack snapping into place.
On the tech side, Google just dropped DiffusionGemma, an open text diffusion model that runs roughly four times faster than earlier baselines. This isn’t a frontier chatbot; it’s an efficiency play that makes high‑quality generative text cheaper and more deployable on ordinary hardware, in tight loop with NVIDIA accelerators. If you want the details, start here: DiffusionGemma launch -> and the company driving it: Google ->.
Building on Tuesday’s Siri overhaul and Wednesday’s Claude Fable 5 launch, the interface story keeps evolving too. Meituan’s new Tabbit AI browser bakes multi‑model agents straight into the desktop, pulling from Chinese labs like DeepSeek and Zhipu AI. So instead of going to a website to talk to an AI, the browser itself becomes the agent shell. You can think of it as a regional twist on the same “agent everywhere” idea: Tabbit AI browser ->.
Meanwhile, Decart’s Oasis 3 world model pushes hard in another direction: photorealistic driving simulations that use huge generative models as synthetic worlds. That tightly couples NVIDIA‑class compute, cloud partners like Amazon and Google, and real‑world autonomy benchmarks. More on that here: Oasis 3 world model ->.
Here’s the thing: as these systems get faster and more embedded, the social and business fallout is now getting its own funding line. Anthropic is setting up a $200M fund to study AI‑driven job disruption, a very direct answer to the “Mythos‑class for the masses” theme from yesterday: Anthropic job disruption fund ->. And Poetic just raised $50M to push reliable enterprise automation built on partners like OpenAI, tightening the link between labs and office workflows: OpenAI ->.
Markets, meanwhile, took a breather. ARM, Qualcomm, Salesforce, Apple, and Palantir all traded down, even as the tech story kept accelerating. If you want to see why investors still crowd around NVIDIA despite the bumps, the Groq deal narrative is a good starting point: NVIDIA–Groq strategy narrative ->. The message from today: the model war is no longer just about bigger LLMs. It’s about faster open models, agentic browsers, simulation worlds, and whether society can keep up.
Get This Delivered Daily
Join thousands of AI professionals who start their day with Race to AGI.