Tencent has released HY-World 2.0, an open-source multi-modal world model that generates navigable 3D environments from text prompts, single images, multi-view images, or video. The model produces editable 3D assets including meshes and 3D Gaussian Splattings that can be directly imported into game engines like Unity and Unreal Engine.
April 16, 2026
One week after Anthropic launched Project Glasswing to let 50+ organizations test its Claude Mythos vulnerability-finding model, the actual CVE count remains unknown. VulnCheck researcher Patrick Garrity found approximately 40 CVEs credited to Anthropic or affiliated researchers since February, but only one—CVE-2026-4747 in FreeBSD—can be directly tied to Glasswing.
GitHub Copilot customers are experiencing rate limits lasting up to 181 hours after the company fixed a token counting bug that had been undercharging for usage of newer models including Claude Opus 4.6 and GPT-5.4. GitHub has suspended Pro free trials and removed Anthropic's Opus 4.6 Fast model from Pro+ tiers.
OpenAI has updated its Agents SDK with sandboxing capabilities that allow AI agents to operate in controlled environments, plus an in-distribution harness for frontier model deployment. The features launch initially in Python, with TypeScript support planned.
Google released a native Gemini application for macOS, marking the company's first standalone desktop client for its AI assistant. The app brings Gemini functionality directly to Mac users without requiring a web browser.
Google released a native Gemini app for Mac on April 15, 2026, joining OpenAI and Anthropic in offering a desktop application. The app supports a global keyboard shortcut (Option + Space), screen sharing with context awareness, local file access, and generation capabilities through Imagen 3 and Veo.
OpenAI has released a major update to its Agents SDK that adds native sandbox support for running AI agents in isolated environments. The framework bundles Model Context Protocol tool usage, code execution, file editing, and custom instructions, with support for sandbox providers including Cloudflare, Vercel, E2B, and Modal.
Meta has released Llama Guard 4, a 12-billion parameter content safety classifier derived from Llama 4 Scout. The model features a 163,840 token context window and can classify both text and image content, available free through OpenRouter with an August 31, 2024 knowledge cutoff.
Google released Gemini 3.1 Flash TTS, a text-to-speech model that accepts detailed prompts to control voice characteristics, speaking style, accent, and delivery. The model is available through the standard Gemini API using the model ID 'gemini-3.1-flash-tts-preview'.
Google released a native macOS app for its Gemini AI assistant, one day after launching a Windows version. The app supports keyboard shortcuts for quick access, screen sharing with files and documents, and image/video generation capabilities.
Anthropic researchers demonstrated that student LLMs inherit undesirable traits from teacher models through distillation, even when those traits are removed from training data. In experiments using GPT-4.1 nano, student models exhibited teacher preferences at rates above 60%, up from 12% baseline, despite semantic screening.
Google has launched a native Gemini app for macOS 15 and later, accessible via Option+Space keyboard shortcut. The app includes screen sharing, local file uploads, and integration with Drive, Photos, and NotebookLM.
Google DeepMind launched Gemini 3.1 Flash TTS, a text-to-speech model that achieved an Elo score of 1,211 on the Artificial Analysis TTS leaderboard. The model introduces audio tags that allow developers to control vocal style, pace, and delivery through natural language commands embedded in text input, with support for 70+ languages.