Weekly AI #23

Welcome to Weekly AI #23.
In 60 Minutes interview, Google DeepMind CEO Demis Hassabis revealed that AGI (Artificial General Intelligence)—AI matching human versatility—could arrive within 2030-2035. He also said that most diseases could be cured in the next decade thanks to AI.

Top 5 Tools

Checklist GG - Finally, get things right

Whisk by Google - Visualize and remix ideas using images and AI

AINavAI - AI Tools List

Dreamina - All-in-one AI creative suite for all your artistic work

ApplyIQ - Get hired smarter—with your personal AI job search agent

Top News/Shorts

Mon, Apr 21
- NariLabs open-sources Dia TTS model with emotional voice cloning
- Windsurf simplifies pricing and extends free GPT-4.1 access
- ComfyUI adds Alibaba's Wan2.1 FLF2V for 720P video generation
- Bytedance launches top-tier AI image model Seedream 3.0
- PlayAI launches instant voice changer with 10-second audio samples
- Together AI speeds diffusion transformers 3.7x with dynamic sparsity
- Anthropic shares Claude Code best practices for LLM development
- Anthropic studies real-world AI values in Claude conversations
- ElevenLabs enables AI call transfers between specialized team agents
- Instagram tests AI to detect teens using adult accounts
- Vidu AI launches Q1 model with advanced video creation features
- Sand.ai releases Magi-1 autoregressive video model open-source
- LTX Studio offers $300 Veo 2 credits for yearly upgrades
- DeepMind CEO predicts AI will cure all diseases within a decade
- AI grenade drone operates autonomously with lethal payload capacity
- Exa AI launches X(formerly Twitter) MCP tool bypassing official API
- Hassabis: AI may gain self-awareness implicitly within 5-10 years
- Skywork AI releases SkyReels-V2 for unlimited cinematic videos

Tue, Apr 22
- Flora introduces image and video upscaling powered by TopazLabs
- Kortix AI launches open-source human-like general AI agent
- Microsoft launches DxGPT AI tool for rare disease diagnosis
- Hedra adds HD characters, voices and AI models in upgrade
- Cluely launches invisible AI desktop assistant with screen and voice monitoring
- Dreamina AI launches Bytedance's Seedream 3.0 for cinematic images
- Suno adds looping video art feature for song customization
- Genspark launches AI Slides for instant presentation creation
- Alibaba Wan offers free Relax Mode or paid priority processing
- Quest 3 uses Stable Diffusion to transform real-world views via mixed reality
- HiDream releases LoRA fine-tuning for custom image generation
- Anthropic to deploy autonomous AI employees by 2026
- Luma AI launches API for Claude and MCP image and video generation
- Tripo open-sources HoloPart for editable 3D shape generation
- Hinton says brains and AI think fundamentally differently
- Gemini Canvas adds LaTeX support for document creation

Wed, Apr 23
- TextArena launches 57+ competitive text games to test LLM skills
- Lovable 2.0 is on the way
- Flora launches Background Remover to keep subject drop rest
- Recraft launches Magic Wand tool for effortless image editing
- Poe launches GPT image generation with advanced features
- Gamma integrates OpenAI's GPT-4o Image for customers
- Nvidia launches Describe Anything for precise image and video captioning
- Microsoft 365 Copilot adds AI agents, notebooks, and cross-app search
- OpenAI launches GPT-Image-1 API for developer image generation
- LiveCC video LLM launches with real-time speech transcription
- Blockade Labs previews instant image-to-skybox style transfer
- Google DeepMind simulates fruit fly behavior with AI model
- Listen Labs raises $27M for AI-powered customer research automation
- Perplexity adds voice assistant for iOS with multi-app control
- Freepik AI Suite creates full fashion campaigns from sketches
- Hunyuan 3D v2.5 enables text and image-to-3D with pro animation tools
- DeepMind CEO envisions AGI-driven best-case future in 10 years
- Nim adds Topaz Upscale for AI-powered video enhancement
- ChatGPT increases o3 and o4-mini limits to 50 messages daily
- Manus offers double credits to early subscribers for one month
- Hailuo AI adds Character Reference for dynamic image transformation
- Ideogram eliminates wait times with Fast and Slow queues
- Ostris launches Flex.2-preview with unified image control features

Thu, Apr 24
- OpenAI increases deep research limits for all user tiers
- Higgsfield integrates OpenAI image model with 30+ animation styles
- Lovable 2.0 launches with multiplayer and enhanced security
- Windsurf plans major platform overhaul within a year
- Hao AI Labs tests show o3 dominates gaming benchmark performance
- Anthropic enhances Claude Code with faster and visual workflow tools
- Google AI launches ZAPBench for whole-brain activity modeling
- Luma AI brings Ray2 video generation to Adobe Firefly
- Perplexity now pre-installed on Motorola phones with Pro trial
- Tavus Hummingbird Lipsync launches on FAL for AI video projects
- Pika Labs launches video manipulation tool "Pikaffects"
- Freepik AI Suite integrates GPT-4o for seamless content creation
- Flora upgrades Intelligent Canvas with GPT-4o integration
- Rime Labs launches Arcana for ultra-realistic AI voice generation
- Google DeepMind unveils Lyria 2-powered Music AI Sandbox tools
- Krea AI integrates ChatGPT Image for multi-style AI generation
- FLock and Alibaba Qwen team up to revolutionize decentralized AI
- Google AI introduces Mobility AI for data-driven transport solutions
- Higgsfield AI launches faster cheaper Turbo model with new motion styles
- Exa Labs announces Company Researcher MCP Tool for insights

Fri, Apr 25
- Docker adds local private LLM support for VSCode and IDEs
- LM Studio 0.3.15 released with new features and RTX 50-series support
- Cognition Labs launches DeepWiki—like Deep Research for GitHub repos
- Google Workspace updates Canvas in Gemini for collaborative creation
- Lemon Slice Live enables real-time AI video chats from photos
- Kimi AI launches open-source audio model with 13M-hour training
- Simular launches as first local macOS AI web-browsing agent
- Manus secures Benchmark funding at $500M valuation
- Alibaba Qwen Chat app launches free on iOS and Android
- Adobe announces mobile app with latest Firefly Image Model 4 and Firefly Video Model
- Baidu launches cost-efficient ERNIE X1 Turbo and 4.5 Turbo models
- Karpathy shares defensive workflow for professional AI-assisted coding

Sat, Apr 26
- RunwayML previews Gen4 References for seamless video consistency
- Leaked specs suggest DeepSeek R2 outperforms GPT-4o at 97% lower cost
- PixVerse App 2.0 launches with AI dance and style features
- Microsoft Copilot integrates with Motorola devices in AI partnership
- GPT-4o update enhances both IQ and personality traits
- Altman's OpenAI accused of abandoning nonprofit AGI safety pledges
- Step1X-Edit launches on Fal for pro-level photo transformations

Sun, Apr 27
- Tripo Doodle open-sourced for real-time 3D asset creation