Linked and quoted from the labs, the press and the people using them. Times in UTC.
The 17 labs we track and the biggest stories beyond them. The rest is in Everything.
Amazon announced new tools for independent sellers at its Accelerate seller conference, including a plugin that lets them run their businesses via AI assistants. The tools also include automations to help sellers.
Meta is building a dedicated hardware device for its new Muse AI agent, called Muse Charm. CEO Mark Zuckerberg briefly showed the product at the end of his Meta Connect presentation.
Meta announced a camera-free version of its AI glasses. The company says they will be lighter and offer up to 12 hours of battery life.
Meta is holding its annual September product launch event in Menlo Park. Mark Zuckerberg's keynote is described as being about how Meta is building a future for everyone.
Mercury 2.5, an LLM, is reported to reach 770 tokens per second on Artificial Analysis. The item links to its model page on the benchmarking site.
Z.ai's GLM-5.3-Prime, a high-speed variant of GLM-5.3, is listed on OpenRouter. It offers 1.5 to 2 times the output throughput and supports text input and output with a 1M-token context.
Anthropic's Claude Devs account said cloud sessions for Claude Code are officially available and out of research preview, letting Claude Code keep working when a laptop is closed. Existing subscribers get a one-time credit of $100 on Pro and $250 on Max, claimable until Oct 7, and GitHub must be connected to start a…
Qwen3.8 Max Prime, a higher-throughput variant of Qwen3.8 Max from Alibaba's Qwen team, is listed on OpenRouter as a separate SKU at a higher price point. It accepts text, image, and video input.
Black Forest Labs released FLUX 3 Action, a 7B model for robotics, with a model page and Hugging Face collection.
YouTube is adding AI tools to its creator studio, including a storytelling assistant that analyzes scripts and rough cuts. Gemini becomes a chat-based editing assistant for Shorts, and a live translation feature turns English streams into Spanish.
Google introduced private, server-side memory to its Private AI Compute offering for personal AI. The feature is described as secure and private.
Google announced Gemini 3.8 text-to-speech on its DeepMind site. The item gives no further details.
Nvidia published Nemotron 3 Diarization, a streaming speaker diarization model that labels up to eight speakers. It has day-zero Transformers integration and was tested by a user in a local voice agent setup.
YouTube introduced custom feeds that let users describe the videos they want to see in their own words. Gemini then builds a personalized feed around that request.
ZeroDrift Inc. launched Anchor 3.0, a family of small language models that check messages generated by AI agents before they are sent. The company said the models enforce financial regulations and internal corporate policies while running fast enough to examine every outgoing message.
NextLM Inc. is bringing its AI Prospecting Agent to Google Cloud Marketplace and Gemini Enterprise. The agent uses behavioral analysis and custom AI models to help sales professionals identify and rank leads.
Alibaba's Qwen team released Qwen-Audio-3.1, a lineup of five models for speech recognition, text-to-speech, and real-time interaction. The company also cut AI audio prices by up to 95 percent.
Meta announced Meta VR Glasses, a virtual reality device in a glasses form factor weighing 100 grams. The company says the device can be worn comfortably for hours.
Meta introduced Ray-Ban Meta Audio, its first audio glasses, at Connect. The company also announced its biggest expansion of AI glasses styles.
Upstage's Solar Mini 4 is now listed on OpenRouter. It is a 35B-parameter mixture-of-experts model with 3B active parameters and a 524K context window, priced at $0.05 per million input tokens and $0.20 per million output tokens.
ChatGPT Ads is expanding to Southeast Asia and Taiwan. The ads product now reaches businesses in more than 60 countries.
OpenAI and Grab launched GO Forward with AI, a regional programme to help 30,000 partners build practical AI skills across Southeast Asia.
Anthropic launched Claude Marketplace, a place to discover plugins, agents, and services from its partners.
AntLing released the open-weight Ming-Image-0.1-Design and Ming-Image-0.1-Design-Layer models, both 6B, plus two open-source Agent Skills for UI design and image-to-editable-PPT. Ming-Image-0.1-Design ranks first among open-weight models on Artificial Analysis's UI/UX Design leaderboard.
Cohere's flagship model Command A+ is now listed on OpenRouter. It accepts text and image inputs with a 192K context window and supports native tool calling with strict tool schemas.