Weekly AI #22

Welcome to Weekly AI #22
OpenAI expands its models with the 4.1 Series (featuring enhanced reasoning and safety guardrails), the lightning-fast o3-mini and o4-mini models for cost-efficient deployment, and the Codex CLI – bringing AI-powered coding directly to developer terminals.

Top 5 Tools

AI Ghibli Video Generator - AI Ghibli video generator is a creative tool that instantly transforms your photos into dreamy

Gamma - Create with AI

Higgsfield AI - The ultimate AI-powered camera control for creators

Jimeng AI - One-stop AI creation plaform

Appsmith Agents - Bring AI agents to work

Top News

OpenAI Launches New Wave of Releases—GPT-4.1 Series, o3 & o4-mini, and Codex CLI - OpenAI just released a wave of new AI tools, including the GPT-4.1 series (faster and cheaper for developers), o3 & o4-mini (multimodal models for coding and visuals), and Codex CLI (AI coding in your terminal). Plus, they're reportedly in talks for a $3B acquisition of Windsurf (ex-Codeium) to boost AI-powered development.

Shorts

Mon, Apr 14
- Anthropic tests Claude voice mode with three voices on iOS
- Neo Gamma robot delivers a TEDx talk, then shows off its chore skills
- alphaXiv adds GPT-4.1 for interactive arXiv paper analysis
- Microsoft Azure adds GPT-4.1 series to OpenAI service
- Microsoft Edge adds free Copilot Vision with screen narration
- Cursor offers free GPT-4.1 access for limited time
- CodeGPT introduces Auto Documentation via codebase knowledge graphs
- Windsurf AI offers free GPT-4.1 access for one week
- OpenAI releases enhanced GPT-4.1 series with million-token context
- Nala Robotics unveils AI-powered Sandwich BOT for automated meals
- ElevenLabs expands AI voice tech with Japan subsidiary launch
- Topaz AI Video Upscaler delivers 4K 60FPS with Freepik AI
- Grok-3 Mini aces hardcore reasoning tests with perfect scores
- Kimi.ai unveils Kimina-Prover theorem prover with SOTA results
- Freepik launches Composition Reference for draft-to-image generation
- OpenAI to introduce new feature in latest live broadcast
- Hugging Face acquires Pollen Robotics for open-source AI robots
- Google DeepMind unveils DolphinGemma for dolphin communication research
- Netflix tests OpenAI-powered AI search for iOS users
- Hunyuan introduces Sonic for global audio-driven portrait animation
- Kling 2.0 coming with major AI enhancements soon
- CodeLLM brings AI coding to VS Code with multi-model smarts
- DeepSeek open-sources inference engine to enhance AI collaboration
- Chinese developer creates firefighter robot from Unitree G1 robot
- MiniMax MCP tools now enable voice and video creation
- ByteDance's Seaweed-7B delivers efficient high-quality video generation foundation

Tue, Apr 15
- OpenAI updates safety framework for advanced AI risk mitigation
- Higgsfield AI unveils Pulse with iconic human motion captures
- Anthropic launches Claude Research with Google Workspace integration
- Google launches Veo 2 on Gemini and Whisk
- Google One AI subscribers get global Whisk Animate access
- WorkflowAI enables rapid AI feature development in minutes
- Gemma app unveils major redesign for creative sharing
- Runway brings Gen-4 to iOS for mobile generation
- Cohere's Embed 4 cuts storage costs by 83% for AI apps
- Microsoft unveils BitNet 2B, first native 1-bit AI model
- Hunyuan introduces Portrait with implicit control technology
- Kling AI launches 2.0 with enhanced video generation
- Fal AI integrates Kling 2.0 video generation via API
- Google DeepMind AI creates superior reinforcement learning algorithms

Wed, Apr 16
- Mistral AI launches customizable classifier builder for developers
- Cursor adds o3 and free o4-mini for limited time
- OpenAI API now offers powerful o3 and efficient o4-mini
- OpenAI negotiates $3B Windsurf acquisition for coding expansion
- Windsurf AI offers free o4-mini access until April 21
- VS Code adds OpenAI's o3 and o4-mini models
- OpenAI releases Codex CLI for local code generation
- OpenAI unveils o3 & o4-mini featuring multimodal tool use
- GitHub Copilot now integrates OpenAI's o3 and o4-mini models
- Fal AI partners with Hedra Labs for new collaboration
- Veed.io enhances text-to-video editing with Cursor-like feature
- Scout AI unveils defense robotics model Fury with $15M funding
- Krisp AI now removes accents from voice calls
- Captions launches Mirage Edit for AI-generated talking videos
- Krea AI integrates Kling 2 for hyper-realistic video generation
- xAI Grok Studio enables real-time coding and document collaboration
- ChatGPT adds Image Library for AI-generated content management

Thu, Apr 17
- Flora launches Gemini-powered image blending for creative designs
- Sync launches AI lipsync studio to streamline video editing
- 1,000 AI agents built a working civilization in Minecraft
- Google launches efficient Gemini 2.5 Flash AI model
- Fal AI offers $20 coupon for first 2000 video model users
- Meta FAIR releases perception AI tools and 8B model weights
- Google demos Android XR AI smart glasses prototype at TED 2025
- OpenAI pursued Cursor twice before settling on Windsurf
- Alibaba launches open-source 14B-parameter Wan2.1 for AI video generation
- Shopdev AI launches Shopify developer tool for custom themes
- Arcads AI enables emotion-controlled avatar generation for videos
- Grok adds memory for personalized conversation responses
- Azure adds new o3 and o4-mini models to OpenAI Service

Fri, Apr 18
- xAI launches Grok 3 family with cost-efficient reasoning models
- Abacus AI launches DeepAgent for automated coding and workflows
- Bill Gates predicts AI may replace doctors and teachers
- ChatGPT combines memory and web search for personalized results
- Google offers free Gemini Advanced for US students until 2026
- Google releases Gemma 3 QAT models for local GPU deployment
- Luma AI adds Cinematic Camera Angle Concepts in DreamMachine
- ChatGPT sets usage limits for o3 and o4-mini models
- Krea AI launches Stage for text-to-3D environment creation
- SkyReels launches first fully automated text-to-film AI agent
- Google's memory-efficient Gemma 3 QAT models hit LM Studio
- Vidu AI teases cinematic animation upgrade launching soon
- Hunyuan open-sources InstantCharacter for customizable AI avatars
- Midjourney adds layered image editing with smart tools
- Midjourney upgrades image editor with layers and smart tools
- FramePack generates unlimited-length videos on low VRAM
- ComfyUI integrates Vivago AI's HiDream-l1 17B model

Sat, Apr 19
- Grok unveils Voice with Vision multimodal AI feature
- Thailand deploys AI Police Cyborg 1.0 with facial recognition
- Beijing hosts world's first humanoid robot half-marathon race

Sun, Apr 20
- Samsung's AI home robot Ballie, which uses Gemini AI launches this summer in US and South Korea
- LTX Studio adds Veo 2 for flexible AI video generation