Xiaomi releases two open source AI models optimized for agentic tasks

Xiaomi has launched two new open source artificial intelligence models: MiMo-V2.5 and MiMo-V2.5-Pro. Carl Franzen reports for VentureBeat that both models are available under the permissive MIT License, meaning businesses and developers can use, modify, and deploy them commercially without restrictions. The models are designed for so-called agentic tasks, where an AI system autonomously completes …

Read more

China orders Meta to unwind $2 billion Manus acquisition

China’s top economic planning body, the National Development and Reform Commission (NDRC), has ordered Meta to reverse its acquisition of AI startup Manus. The deal was valued at over $2 billion when Meta completed it in late 2025. The NDRC said it would “prohibit foreign investment in Manus in accordance with laws and regulations” and …

Read more

Copilot agentic features now standard in Word, Excel, and PowerPoint

Microsoft has expanded its artificial intelligence assistant Copilot with new capabilities that allow it to take direct, multi-step actions inside Word, Excel, and PowerPoint. Sumit Chauhan, President of the Office Product Group, writes that these so-called agentic features are now generally available and the default experience for subscribers. Previously, Copilot could answer questions but struggled …

Read more

Tencent launches international beta for AI agent tool QClaw

Tencent has opened an international beta for QClaw, an artificial intelligence agent product built by its PC Manager team. T. K. Lin reports for KrASIA that the tool is designed for non-technical users and allows them to deploy and operate AI agents through messaging apps such as WhatsApp or Telegram, without any terminal commands or …

Read more

OpenAI launches workspace agents for business teams

OpenAI has introduced workspace agents in ChatGPT, a new product that lets business teams build and share AI agents capable of handling complex, multi-step tasks across workplace tools like Slack, Google Drive, Salesforce, and Microsoft apps. Workspace agents are powered by Codex, OpenAI’s cloud-based AI coding platform. Unlike previous chat-based interactions, these agents can write …

Read more

OpenAI releases GPT-5.5 with improved agentic capabilities

OpenAI has announced GPT-5.5, its latest large language model, positioned as a significant step forward in handling complex, multi-step tasks on a computer. The model is rolling out to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex. According to OpenAI, GPT-5.5 is designed to handle “agentic” work: tasks where an AI model operates …

Read more

Perplexity launches AI agent “Personal Computer” for Mac

Perplexity has launched Personal Computer, a new feature that brings its AI agent software directly to Mac computers. The tool is an expansion of Perplexity Computer, which the company describes as an all-in-one digital worker capable of creating and executing entire workflows. Personal Computer is now rolling out to Perplexity Max subscribers, who pay $200 …

Read more

Perplexity’s revenue surges 50% as it shifts focus to AI agents

Perplexity has recorded a 50 percent revenue jump in a single month, driven by its expansion into AI agents. Cristina Criddle reports for the Financial Times that the company’s estimated annual recurring revenue (ARR) climbed to over $450 million in March, following the launch of a new agent tool and a switch to usage-based pricing. …

Read more

Anthropic expands Claude Cowork with enterprise controls

Anthropic is rolling out a set of organizational controls for Claude Cowork, its AI-powered collaboration tool, making it available across all paid plans. A post in the official Claude Blog states that the update brings role-based access controls, group spending limits, expanded observability, and new usage analytics to help companies deploy the tool company-wide. Claude …

Read more

Google releases Gemma 4, its most capable open AI model family

Google has launched Gemma 4, a new family of open-weight AI models that the company describes as its most capable to date. The models are built on the same research and technology as Google DeepMind’s proprietary Gemini 3 system and are released under an Apache 2.0 open-source license, which allows developers to use and modify …

Read more

×