Other labVoice & audio
UnconfirmedNotableOpenFish Audio S2 Pro
Fish Audio, a voice technology maker, may be releasing an S2 Pro speech model that could be free to download and run on your own computer.
News · earlier edition
Nvidia dominated the day after CEO Jensen Huang resigned amid a China chip-sales and spy scandal, with Treasury demanding answers and Congress weighing tighter export rules. On the model side, OpenAI said GPT-6 Astra helped produce a proof on a lattice Standard Model definition, and Meituan released a 1.6-trillion-parameter agent model trained on Chinese hardware.
42 stories ranked · updated Nov 15, 23:02 UTC
Huang stepped down over relationships, including one with an executive at a company suspected of funneling chips to China. Treasury Secretary Scott Bessent wants Nvidia to hire an outside CEO and answer within 10 days whether it pushed seemingly illicit chip sales to China.
OpenAI published a candidate nonperturbative lattice definition of the Standard Model and said its GPT-6 Astra model helped develop it into a detailed proof. OpenAI also launched a science blog sharing research methods and researcher accounts.
The LongCat-2.5 model was trained entirely on Chinese hardware and scores 88.4 on Terminal Bench, just behind Claude Opus 5, according to The Decoder.
Rather than computing pixels, the Language model (LLM): A model trained on huge amounts of text to read and write. The engine behind chatbots like ChatGPT, Claude and Gemini. writes Python code that places hundreds of thousands of simulated brushstrokes on a virtual canvas.
Democrats and Republicans have voiced support for further restricting AI chip exports after the Nvidia scandal. The US and Pax Silica allies, including Japan, the UK and India, agreed to create a joint list of AI technologies and supply chains to protect from China.
Bloomberg reports, citing sources, that Apple will focus heavily on integrating third-party iPhone apps with its Siri AI in the glasses.
Epoch AI analysis finds Huawei's Horinger, among China's largest AI data centers, has roughly 7% of the compute of SpaceX's Colossus 2 and may stay far behind until at least 2028.
Releases
Other labVoice & audio
UnconfirmedNotableOpenFish Audio, a voice technology maker, may be releasing an S2 Pro speech model that could be free to download and run on your own computer.
Meituan, the Chinese delivery and services company, released a free-to-download tool that makes realistic talking-person videos from audio, useful for creators building avatar videos.
MiniMaxText & chat
UnconfirmedNotableOpenA possible new chat model from MiniMax, spotted before launch, that developers may be able to run on their own computers for chat and coding.
Turn on the daily digest for any channel: email, Telegram, Discord, push or ntfy.
Archive