Weekly AI #16

Welcome to Weekly AI #16.
This week, Manus AI Agent takes center stage, showcasing its ability to automate complex workflows and bridge the gap between human intent and machine execution.

Top 5 Tools

Omagic AI - Omagic AI is a groundbreaking AI-powered creative platform that democratizes professional

Channel42 - A video editor like Capcut that lets you use the latest AI video models all in one place

Kive Product Shots - AI Product Shots

Spark-TTS - An efficient text-to-speech model using single-stream speech tokens

GrowthTrack AI - Get AI-powered insights to boost your online revenue. Simple, actionable, and completely free

Top News

Worlds First Fully Autonomous AI Agent, Manus - Manus, an AI developed in China, operates independently with no human input, revolutionizing industries and challenging Western tech dominance. Its multi-agent architecture enables unprecedented task automation, posing both innovative possibilities and ethical dilemmas in the AI realm.

Shorts

Mon, Mar 3
- Microsoft releases DeepSeek R1 models for Copilot+ PCs via Azure AI
- Grok accurately predicts all Oscar winners this year
- LM Studio launches SDK with autonomous task execution API
- Lovable launches Versioning 2.0 for easier app version restoration
- ElevenLabs extends Scribe's audio limit to 2 hours with diarization
- Meta AI researchers develop non-invasive brain-to-text decoding method
- Google Colab launches Data Science Agent powered by Gemini
- Anthropic raises 3.5B in Series E funding, valued at 61.5B
- Ray2 now on Poe, transforming images into realistic videos
- Microsoft launches Dragon Copilot to ease healthcare admin tasks
- Deutsche Telekom and Perplexity to launch AI Phone under $1000 in 2026
- Stability AI launches Stable Audio Open for offline mobile audio generation
- Opera launches AI-powered Browser Operator for task automation
- Minimax AI launches Image-01, a cost-effective text-to-image model

Tue, Mar 4
- Fal AI adds DiffRhythm, it can generate full song under-10-second
- Play AI introduces advanced voice AI with human-like interactions
- HeyGen simplifies UGC ad creation with AI avatars and scalability
- OpenAI launches NextGenAI consortium with $50M for AI research
- Cohere AI launches Aya Vision models for multilingual AI tasks
- Hailuo AI launches video app for enhanced content creation
- Hunyuan I2V open-source release coming soon on March 6th
- ElevenLabs partners with Deutsche Telekom for AI-powered podcast creation
- DiffRhythm launches open-weight model for instant song generation
- Unitree's G1 robot showcases advanced martial arts and agility
- Kling AI launches Android app with powerful creative tools
- Cogview4 6B open-source image model released with 2048 resolution
- Pixverse introduces 360 microwave template for consistent characters

Wed, Mar 5
- OpenAI outlines safety and alignment principles for AGI development
- Windsurf Wave 4 update brings new features and improvements
- Google previews Whisk Animate for turning images into 8-second clips
- Qwen's QwQ-32B model now available in LM Studio
- ChatGPT 4.5 rollout to Plus users now complete
- Alibaba officially releases QwQ-32B, a competitive reasoning AI model
- Perplexity adds voice mode to macOS app for seamless interaction
- ChatGPT 4.5 rolls out to Plus users over the next 1-3 days
- Luma AI adds Extend & Loop features to Ray2 for enhanced video creation
- Google introduces AI Mode in Search for advanced query handling
- Fal AI launches eye correction model for video and avatar editing
- OpenAI plans tiered AI agent pricing from 2K to 20K monthly
- LTX Video updates with commercial license, keyframes, and video extension
- Manus launches as the first general AI agent for collaboration
- Rendernet adds AI characters to real products for stunning visuals
- Spark-TTS launches LLM-powered text-to-speech model with voice cloning
- Runway announces Video Restyle feature coming soon for creative editing
- Grok Voice now available to all users on the Grok app

Thu, Mar 6
- ElevenLabs adds voice speed controls for precise dialogue pacing
- Nim integrates LTX video generation tools for all users
- Mintlify launches MCP generator for easy API and content integration
- Google AI launches AMIE for advanced medical diagnosis and treatment
- Fal AI integrates Hunyuan i2v for advanced video generation
- Lovable introduces dev-mode for direct code editing in early access
- Fal AI partners with RunDiffusion to host Juggernaut AI models
- ChatGPT for macOS now edits code in IDEs for Plus, Pro, and Team users
- Tavus launches Phoenix-3, Raven-0, and Sparrow-0 for advanced AI interaction
- Hedra Labs launches Character-3 and Hedra Studio for AI video creation
- GitHub Copilot updates with Jupyter edits, agent mode UX, and custom instructions
- Anthropic updates Console for collaborative prompt building with Claude
- Mistral AI launches Mistral OCR for advanced document understanding
- Pika Labs introduces 1080p Pikadditions and Pikaswaps for paid users
- Alibaba's Wan invites users to start creative video adventures on their website
- Freepik introduces Colors for custom AI image generation with color palettes
- Kling AI introduces AI Effects for creative image-to-video transformations
- Krea AI integrates Pika's new features for enhanced creativity
- Model Context Protocol (MCP) emerges as a new AI interaction standard
- Convergence AI launches Template Hub to automate global tasks
- Qwen adds QwQ-32B to Qwen Chat for direct user access
- Tencent releases HunyuanVideo I2V for transforming images into videos
- Zyphra launches Zonos in Japanese with expressive language capabilities
- Channel42 public beta launches with AI video generations and video editing tools

Fri, Mar 7
- ElevenLabs partners with Google Cloud to scale voice AI for enterprises
- Apple postpones AI-powered Siri update to 2026 due to development challenges
- Google AI launches experimental Gemini Embedding text model with 8K token limit
- Luma AI launches Ray2 Flash for faster, cheaper video generation in high quality
- Reflection AI, founded by former Google team, launches with $130M for autonomous coding systems
- OpenAI introduces new models page for detailed capability comparisons
- Microsoft tests AI models from xAI, Meta, and others to diversify Copilot
- Grok now responds to user queries in X replies—just tag @grok and ask
- Windsurf AI simplifies MCP integration for seamless tool connectivity
- Midjourney begins closed testing of v7 model with fine-tuning
- Microsoft enhances Think Deeper with o3-mini for advanced queries
- Windsurf adds Browser Previews and component mentions for Cascade