Sponsored by
Good morning, {{first_name | AI enthusiast}}.
Black Forest Labs, the startup behind FLUX, just extended its video-and-image model into the physical world. A new system called FLUX-mimic uses that same visual model to control robots — and Audi is already running it on real production lines.
The model reportedly needs just 30 minutes of demonstration footage to pick up a new task, a fraction of the time older robotics systems require. If a model trained mostly on video and images can generalize to physical manipulation this fast, how much longer until “foundation model” stops meaning just chatbots and image generators?
Today in AI Brief:
Black Forest Labs’ FLUX 3 learns to run robots
Stripe eyes a $10 billion bet on OpenRouter
OpenAI brings ChatGPT Work to small businesses
The integrated coworker for teams
Adapt brings frontier AI to your team in Slack. Tag @Adapt to answer a question, run a task on a schedule, or build the dashboard you've been waiting weeks for. It connects to your tools, learns your business, and does real work.
Black Forest Labs’ FLUX 3 Learns to Run Robots
In Brief: Black Forest Labs released FLUX 3 in early access, a multimodal model that generates image, video, and audio together — and built a companion system called FLUX-mimic that uses the same backbone to control robots.
The Details:
FLUX 3 generates up to 20 seconds of video with synchronized dialogue, sound effects, and music, with testers preferring its output over Runway Gen-4.5 and Kling v3 Pro in head-to-head comparisons.
FLUX-mimic learns a new task from roughly 30 minutes of demonstration footage instead of the 30-plus hours typically required, decoding actions straight from the model’s video-prediction pathway.
Audi is already running FLUX-mimic on production and logistics lines for tasks like kitting parts and inserting delicate electronic components, with the system reacting in just 101 milliseconds.
Take Away:
Black Forest Labs built its reputation on image generation, but FLUX-mimic turns that same visual model into a controller for physical robots. Betting that one foundation model can both make videos and run a factory floor tests how far “world models” can stretch beyond content creation.
Stripe Closes In on a $10 Billion Bet on AI Infrastructure
In Brief: Stripe is in talks to acquire OpenRouter for roughly $10 billion, the Wall Street Journal reported, a deal that would value the AI model marketplace nearly eight times its $1.3 billion valuation from May.
The Details:
OpenRouter lets developers access and switch between hundreds of AI models from providers like OpenAI and Anthropic without rebuilding their applications.
Stripe is also pursuing a $53 billion joint bid for PayPal with Advent International, signaling a broader push into M&A.
The talks are still ongoing and could fall apart, and other technology companies are reportedly weighing competing bids for the three-year-old startup.
Take Away:
A payments company paying near-$10 billion for an AI routing layer shows how much value is concentrating in the infrastructure that sits between developers and model providers, not just the models themselves. It’s also a sign Stripe wants a stake in AI spending beyond processing the transactions.
One Account. Every Market. No Closing Bell.
Markets don't wait for Monday. News breaks on a Saturday morning, and most traders can do nothing but watch.
Not on Liquid. Trade domestic and international equities, commodities, forex, crypto, and prediction markets — all from one account, 24 hours a day, 365 days a year. Liquid gives you access to any market, from anywhere, anytime. To us, access is arbitrage.
Getting started takes under 10 minutes: log in with Google, deposit with Apple Pay or a bank transfer, and trade from your phone or desktop — wherever you are in the world.
While everyone else is refreshing headlines and waiting for the open, you're already positioned. That's the difference between reacting to markets and actually trading them.
OpenAI Brings ChatGPT Work to Small Businesses
In Brief: OpenAI launched a small business program that pairs ChatGPT Work with hands-on training and app integrations, while announcing that ChatGPT Work and Codex have reached a combined 10 million users.
The Details:
The 10 million combined user count for ChatGPT Work and Codex doubles Codex’s user base from earlier this month, after OpenAI folded the standalone Codex app into ChatGPT’s desktop client.
ChatGPT Work connects to Slack, Google Drive, and Microsoft Teams, letting the GPT-5.6-powered agent maintain context and execute multi-step tasks across a business’s existing tools for hours at a stretch.
OpenAI now counts 5 million-plus business users and over 1 million enterprises using its tools, alongside in-person Small Business AI Academies launching across the U.S.
Take Away:
Small businesses have been slower to adopt agentic AI than enterprises with dedicated IT teams, and OpenAI is betting that free training plus ready-made integrations closes that gap. If ChatGPT Work sticks with small business owners the way it has with Codex, OpenAI adds a large, sticky user base outside the enterprise contracts it already dominates.
Everything else in AI
The Department of Energy selected 278 Genesis Mission research projects worth $293 million, aiming to double U.S. scientific productivity with AI and national-lab supercomputing.
Meta launched a standalone Seller app that uses AI to draft Facebook Marketplace listings — titles, prices, and categories — from a single photo.
Moonshot AI is pursuing a $50 billion valuation in a final pre-IPO funding round, up from $31.5 billion, as its Kimi K3 model drives a revenue surge.
OpenAI rolled out Health in ChatGPT to all U.S. adults, letting users connect Apple Health data and medical records as weekly health queries hit 300 million.