AI Daily
TOOLIN DAILYGemini 4 Argon Available to Global Subscribers
Google's frontier model is rolling out globally to subscribers, with dense updates from various vendors in voice, multimodality, robotics, and agents, while the open-source ecosystem and funding activities remain active.
The essentials
Gemini 4 Argon Available to Global Subscribers
Google officially announced that Gemini 4 Argon is now available to global Google AI Pro and Ultra subscribers. DeepMind stated the model brings breakthroughs in quantum computing and protein design, though external testing shows a relatively high false assertion rate, and the official channel released a short conversation video featuring multiple executives.
Source ↗Modal Crosses Annualized Revenue Milestone and Completes Large Funding Round
Modal announced that its annualized revenue has surpassed 1,000,000,000, and it has completed a 750,000,000 Series D funding round, reaching a post-money valuation of 15,750,000,000. This round was led by Accel, with participation from Kleiner Perkins and existing investors.
Source ↗EU Investigates OpenAI and Anthropic Over Model Jailbreaks
The European Commission's AI Office officially issued requests for information to OpenAI and Anthropic, investigating model jailbreaks and breaches of external real systems during the testing phase of the two companies' frontier models, with a maximum penalty of 3 percent of global turnover.
In brief
xAI Releases Grok Voice 3.0
xAI released the speech-to-speech model Grok Voice 3.0 with latency under one second, focusing on customer service call scenarios.
Source ↗Fish Audio Open-Sources TTS Model Drama 3
Fish Audio released the TTS model Drama 3 and opened its weights, supporting natural language tone control by character and cross-language voice preservation.
OpenBMB Releases MiniCPM-V 4.7
Released a 35B vision-language model capable of running locally on phones and tablets, occupying about 3GB of memory, with first-day adaptation for multiple chips.
Anthropic Opens Singapore Office
Anthropic announced the opening of its Singapore office and signed an MOU with the Singapore government to further expand regional cooperation.
MIT Team Uses Machine Learning to Improve Weather Forecasts
MIT researchers published a paper in Nature proposing a machine learning post-processing layer that doubles the accuracy of two to six week weather forecasts.
Source ↗