Kling AI previews its unified VIDEO 3.0 model, offering longer clips, Multi-Shot storyboard workflow, and stable references in early access.

Claude Sonnet 5 leaks suggest a new Anthropic model with strong math and coding capabilities, launching this week.

What do we know so far? Perplexity may soon introduce Model Council for Max users, allowing multi-model system advancements and hinting at a new ASI mode.

Anthropic is testing upgraded voice mode for the desktop app and a new knowledge base feature for Claude.

Google is testing a new Gemini checkpoint in AI Studio, impressing early users with precise UI generation and SVG output, but release details remain unclear.

Google is testing a “Personal intelligence” layer in NotebookLM, hinting at personalized responses and custom prompts based on chat history.

Notion is testing new agent features, including a redesigned settings UI, new automation triggers, and upcoming Agents 2.0 upgrades.

Meta AI is testing Avocado models, MCP integrations, and Manus browser agent support, with scheduled tasks and OpenClaw compatibility launching soon.

Anthropic is testing a Tasks feature in Claude’s mobile apps, bringing Cowork-style automation, repeatable actions, and possible browser tasks soon.

Perplexity is preparing to launch its Gamma mode, featuring rapid, agent-driven searches powered by Grok 4.1 or possibly Grok 4.20.

xAI's Grok Build is evolving into a full IDE with multi-agent coding, arena mode, dictation, browser-style tabs, and GitHub integration.

Microsoft is testing a Health tab in Copilot with connectors for Fitbit, Garmin, Oura, and Apple Health, centralizing health data in one space.