Discover quality AI tools
Bring together the world’s leading artificial intelligence tools to help improve work efficiency
Articles published by RichardGonzález
Google is developing an experimental AI model named Mathematica, based on DeepThink V3, optimized for complex math and heavy computing. This unstable internal version supports a 1 million token context window with a 65536 token output limit. Labeled Teamfood, it is currently undergoing internal employee testing, following recent releases like Gemini 3.8 Live and Live Extended Thinking.
Google is developing an experimental AI model named Mathematica, based on DeepThink V3, optimized for complex math and heavy computing. This unstable internal version supports a 1 million token context window with a 65536 token output limit. Labeled Teamfood, it is currently undergoing internal employee testing, following recent releases like Gemini 3.8 Live and Live Extended Thinking.
SoftBank completed a $10 billion additional investment in OpenAI through Vision Fund 2, funded by external borrowing. This is the second phase of a $30 billion plan; the third $10 billion phase is expected by October 1. SoftBank previously invested $500 million in OpenAI in 2024.
SoftBank completed a $10 billion additional investment in OpenAI through Vision Fund 2, funded by external borrowing. This is the second phase of a $30 billion plan; the third $10 billion phase is expected by October 1. SoftBank previously invested $500 million in OpenAI in 2024.
DuckDuckGo sees a major US traffic surge as users flee Google's AI-powered search. App installs jumped 18% weekly, with iOS up 33%, following Google's AI push without a clear opt-out. Visits to DuckDuckGo's AI-free search page also rose 23%, highlighting demand for user control and privacy over automated results.
DuckDuckGo sees a major US traffic surge as users flee Google's AI-powered search. App installs jumped 18% weekly, with iOS up 33%, following Google's AI push without a clear opt-out. Visits to DuckDuckGo's AI-free search page also rose 23%, highlighting demand for user control and privacy over automated results.
Kuaishou plans to spin off its video generation model Kuaishou AI for a Pre-IPO round targeting a 20 billion USD valuation aiming for a listing next year The unit is already profitable with 500 million USD in annual recurring revenue and has attracted investor interest including Tencent amid fierce competition in Chinas AI sector
Kuaishou plans to spin off its video generation model Kuaishou AI for a Pre-IPO round targeting a 20 billion USD valuation aiming for a listing next year The unit is already profitable with 500 million USD in annual recurring revenue and has attracted investor interest including Tencent amid fierce competition in Chinas AI sector
China Mobile launched the Mobile Model Management Platform (MoMA) at the Mobile Cloud Conference. The platform integrates over 300 mainstream AI models, including its self-developed "Jiu Tian" model, to lower AI application thresholds. It features an intelligent routing engine for model selection, a "Confidential Model" service for data security, and a pay-as-you-go billing model that reduces costs by about 30% per token.
China Mobile launched the Mobile Model Management Platform (MoMA) at the Mobile Cloud Conference. The platform integrates over 300 mainstream AI models, including its self-developed "Jiu Tian" model, to lower AI application thresholds. It features an intelligent routing engine for model selection, a "Confidential Model" service for data security, and a pay-as-you-go billing model that reduces costs by about 30% per token.
Alibaba launches Wan2.7-Image, a unified AI model for image generation and editing. It breaks limitations like generic AI faces by enabling high-precision virtual character customization for unique portraits. The model features a color palette extractor for artistic styles and supports print-quality, multi-language text rendering with complex layouts. It allows interactive pixel-level editing and maintains consistency across up to nine subjects. Built on a unified generation-understanding architecture for semantic comprehension, it is now available via Alibaba Cloud for applications in short dramas, e-commerce, and education.
Alibaba launches Wan2.7-Image, a unified AI model for image generation and editing. It breaks limitations like generic AI faces by enabling high-precision virtual character customization for unique portraits. The model features a color palette extractor for artistic styles and supports print-quality, multi-language text rendering with complex layouts. It allows interactive pixel-level editing and maintains consistency across up to nine subjects. Built on a unified generation-understanding architecture for semantic comprehension, it is now available via Alibaba Cloud for applications in short dramas, e-commerce, and education.
The FTC under the Trump administration removed three blog posts from the Lina Khan era addressing AI risks and open-source AI. This is part of a broader pattern of content removals aligning with a shift towards deregulation and away from previous priorities like AI safety.
The FTC under the Trump administration removed three blog posts from the Lina Khan era addressing AI risks and open-source AI. This is part of a broader pattern of content removals aligning with a shift towards deregulation and away from previous priorities like AI safety.
Mira Murati's Thinking Machines Lab tackles AI model nondeterminism in its research blog, aiming to create reproducible responses by controlling GPU kernel orchestration. This could enhance reliability for enterprises and improve reinforcement learning. Backed by $2B funding, the lab plans to release a product soon for researchers and startups customizing AI models.
Mira Murati's Thinking Machines Lab tackles AI model nondeterminism in its research blog, aiming to create reproducible responses by controlling GPU kernel orchestration. This could enhance reliability for enterprises and improve reinforcement learning. Backed by $2B funding, the lab plans to release a product soon for researchers and startups customizing AI models.
Libby, a library e-book and audiobook app, introduces an AI-powered 'Inspire Me' feature for personalized book recommendations based on user prompts or saved titles. While it prioritizes available library titles and ensures user data privacy, some readers and librarians express concerns over AI integration and prefer human-curated suggestions. The feature, soft-launched this month, will roll out to all users in September.
Libby, a library e-book and audiobook app, introduces an AI-powered 'Inspire Me' feature for personalized book recommendations based on user prompts or saved titles. While it prioritizes available library titles and ensures user data privacy, some readers and librarians express concerns over AI integration and prefer human-curated suggestions. The feature, soft-launched this month, will roll out to all users in September.
