OpenAI has released three new voice models for developers through its Realtime API. Each model focuses on a different capability: reasoning, translation, and transcription. The first model, GPT-Realtime-2, brings GPT-5-class reasoning to live voice conversations. According to OpenAI, it can handle complex requests, manage interruptions, and call external tools while keeping a conversation going naturally. …
Mistral AI has launched Workflows, a production-grade orchestration platform designed to help businesses deploy AI reliably across critical operations. Michael Nuñez reports for VentureBeat that the Paris-based company, valued at approximately $13.8 billion, released the product in public preview as part of its Studio platform. Workflows allows engineers to define, run, and monitor multi-step AI …
Google has introduced two upgraded versions of its autonomous research agents, aiming to turn AI-powered analysis into a core tool for enterprise workflows. The new systems, called Deep Research and Deep Research Max, are built on the Gemini 3.1 Pro model and target tasks ranging from fast summaries to complex, multi-step investigations. The official blog …
Anthropic has released Claude Code Channels, a feature that lets developers send messages to a running Claude Code session directly from Telegram or Discord. A developer can type a prompt on their phone, and Claude executes the task on their desktop or server and replies through the same messaging app. The feature is built on …
OpenAI has released two new AI models, GPT-5.4 mini and GPT-5.4 nano. As the names suggest: Both are smaller and faster versions of the company’s flagship GPT-5.4 model, designed for high-volume tasks where speed and cost matter. According to OpenAI, GPT-5.4 mini runs more than twice as fast as its predecessor, GPT-5 mini, and shows …
Google has released an open-source command-line interface (CLI) that gives AI agents direct access to its Workspace suite, including Gmail, Google Docs, Sheets, Drive and Calendar. Carl Franzen reports for VentureBeat that the tool, called googleworkspace/cli, is designed for both human developers and automated AI agents. The CLI allows developers to interact with all major …
OpenAI has launched GPT-5.3-Codex, a coding model that the company describes as its most capable agentic coding tool to date. The model runs 25% faster than its predecessor and combines advanced coding performance with reasoning capabilities in a single system. According to OpenAI, GPT-5.3-Codex marks a significant milestone as the first model that helped create …
OpenAI has released a desktop application for macOS that transforms its Codex AI coding system into what the company calls a command center for managing multiple autonomous coding agents. The app allows developers to delegate multiple coding tasks simultaneously and supervise AI systems that can run independently for extended periods. The Codex app introduces several …
Google has introduced the Universal Commerce Protocol (UCP), an open standard designed to enable AI agents to handle various aspects of online shopping. The company developed the protocol in collaboration with Shopify, Etsy, Wayfair, Target, and Walmart. The protocol allows AI agents to manage different stages of the customer journey, from product discovery to post-purchase …
Anthropic has released Agent Skills as an open standard, making the technology available across different AI platforms. The company first introduced skills in October as a way to teach Claude repeatable workflows. Now the system works beyond Claude, with Microsoft, Cursor, and other platforms adopting it. Skills are folders containing instructions and scripts that tell …