Moonshot AI, the Beijing based startup backed by Alibaba, has released Kimi K3, a 2.8 trillion parameter model the company calls in an official blog post the world’s first open 3T-class model. The system combines two new architectural components, Kimi Delta Attention and Attention Residuals, with a 1 million token context window and native vision …
Thinking Machines, the AI startup founded by former OpenAI chief technology officer Mira Murati, has released its first major language model under an open source license. Carl Franzen reports for VentureBeat that the model, called Inkling, targets enterprises that want to run AI on their own servers while keeping costs low and avoiding ideological filtering. …
A new open-source framework called Agent-as-a-Router promises to make AI model selection smarter and cheaper. Ben Dickson reports for VentureBeat that the concrete implementation, called ACRouter, beats both static routing systems and the common strategy of defaulting to premium models like Claude Opus. Companies use model routing to send simple tasks to cheap, fast models …
Three major AI developers released new models within days of each other this week, and all of them lead with the same pitch: lower cost per task. Lorelei Smillie and Rachel Metz report for Bloomberg that OpenAI’s GPT-5.6, SpaceXAI’s Grok 4.5 and Meta’s Muse Spark 1.1 all promise more output for fewer tokens, the units …
Slack has launched a major integration that lets its built-in AI assistant, Slackbot, tap directly into Salesforce’s CRM data, Tableau analytics, Data 360 customer profiles and dozens of third-party apps, all from a single chat message. Michael Nuñez reports for VentureBeat that the feature relies on dedicated Model Context Protocol (MCP) servers connecting Slackbot to …
Tencent has released the full version of its Hy3 language model under the permissive Apache 2.0 license, removing regional restrictions that previously excluded the European Union, the United Kingdom and South Korea. Sam Witteveen reports for VentureBeat that the license reversal, more than the model’s capabilities, drove the strongest reaction from the open-model community. Hy3 …
Companies pour billions into artificial intelligence, yet most fail to turn that investment into measurable business value. Colleen Jones writes for Content Science Review that this disconnect, which she calls the “AI strategy gap,” separates organizations that merely adopt AI tools from those that transform how they create value. The numbers support her argument. MIT …
Microsoft is launching a new business unit called Microsoft Frontier Company, backing it with $2.5 billion and 6,000 engineers who will work directly inside client organizations. Judson Althoff, CEO of Microsoft’s commercial business, writes on the Official Microsoft Blog that the initiative goes beyond so-called forward-deployed engineering, aiming to be the largest outcome-driven engineering organization …
Google has released a new AI image generation model aimed at developers and enterprise teams who need speed and low costs above all else. The model is officially called Gemini 3.1 Flash-Lite Image but is marketed under the name Nano Banana 2 Lite. It generates images at 1,000 pixel resolution in around four seconds and …
Google has opened its Gemini Omni Flash model to developers and enterprise customers via API, bringing conversational video editing to professional workflows for the first time. Sam Witteveen reports for VentureBeat that the model, which debuted at Google I/O 2026, lets users generate and refine video clips through plain-language instructions rather than prompt-by-prompt generation from …