Weekly AI #21

Welcome to Weekly AI #21
At Google Cloud Next 2025, the tech giant unveiled an impressive array of AI innovations. They introduced the Ironwood TPU, their fastest AI chip yet, along with Gemini 2.5 Flash, which is crafted for more efficient reasoning. And let's not forget Firebase Studio—a true game-changer for mobile and web developers.

Top 5 Tools

FLORA - Your intelligent canvas

Dreamchanted - Turn boring photos into magical art

Ghibli AI Generator - Ghibli AI generator free, Ghibli style image generator

WanVideo - Create AI videos, images & music in one platform

Easy Comment Generator - Free AI comment generator

Top News

Google Unveils Firebase Studio - Firebase Studio, introduced by Google, is an innovative AI platform designed to make app development easier for developers. It accommodates a variety of programming languages and frameworks, providing ready-made templates and collaborative tools that enable quick custom app creation in a coding environment that feels just like home.

Canva Unveils Major Update - Canva's latest update, Visual Suite 2.0, revolutionizes creativity by merging design, spreadsheets, coding, and photo editing into a unified platform. Featuring Canva Sheets, Magic Studio upgrades, and new AI tools, it promises an effortless creative workflow with seamless integrations and enhanced intuitiveness.

Shorts

Mon, Apr 7
- Amazon Nova Reel 1.1 generates 2-minute style-consistent videos
- "One-Minute Video Generation with Test-Time Training" paper enables AI to generate one-minute coherent videos
- MCP supercharges WhatsApp with AI voice/image processing powers
- Runway's Gen-4 Turbo creates videos 5x faster, half cost
- OuteTTS releases open voice AI models on Hugging Face
- ElevenLabs MCP enables voice agents for tasks via text
- Vogent's AI clones voices like Trump with 200ms speed
- Google AI Mode now answers via text, voice and images
- Poe integrates Kling Pro v1.6 for top-tier image-to-video
- VS Code Agent Mode: MCP-powered AI edits code with extensions support
- Gemini transcribes, summarizes YouTube videos with AI Studio link
- Copilot adds AI podcast creator and content organizer
- Shopify mandates AI integration in all employee workflows
- Gemini Advanced now supports camera sharing on Pixel/Samsung
- ElevenLabs integrates GroqCloud to power voice agents with Llama 4
- Llama 4 Scout/Maverick now run real-time on GroqCloud
- Meta's new Llama 4 AI accused of faking benchmark scores

Tue, Apr 8
- Kairos AI automates workflows by learning tasks just once
- Together AI open-sources DeepCoder-14B coding model and data
- Amazon Nova Sonic launches for natural speech-to-speech AI applications
- Perplexity offers startups $5K API credits and Enterprise Pro
- DeepCogito pursues superintelligence with scaled open AI models
- Google's new AI geospatial tool answers complex planet questions
- Ideogram joins Captions for instant AI image-to-video integration
- heyBossAI builds custom sites/apps in 9 minutes via AI team
- Viggle Mic 2.0: Animate characters with just audio or text
- Anthropic study reveals how students use Claude in education
- Krea AI secures $83M funding from top-tier investors
- Adobe releases ultimate AI video prompt guide with examples
- Fal AI integrates Lipsync-2 for zero-shot lip synchronization
- AlphaXiv Assistant speeds up research with AI literature reviews
- Tencent's ACTalker creates ultra-realistic voice-to-lip-sync animations
- Genspark AI breaks digital-physical barrier with real-world calling
- Llama 4 now available in LM Studio
- Fal AI adds Cassette AI for instant music/sfx generation

Wed, Apr 9
- Students get free SuperGrok access for 2 months
- Suno adds extra credits, listening hub, and heavy metal model
- Orpheus launches open-source multilingual TTS models for global access
- Windsurf Wave 7 brings Cascade to JetBrains IDEs
- Kimi.ai launches open-source VLMs with reasoning at 3B scale
- Cline now works dynamically within your active Chrome browser
- Octave TTS upgrade delivers most natural AI speech yet
- Anthropic's Claude Max Plan offers 20x Pro usage limits
- Google DeepMind's Lyria creates custom soundtracks in minutes via AI
- Google's Agent Development Kit simplifies multi-agent AI creation
- Google's Firebase Studio builds apps via natural language commands
- Google AI Studio's redesign earns praise for intuitive interface
- Google's A2A Protocol standardizes AI agent interoperability industry-wide
- WordPress launches free AI builder for no-code websites
- LTX Studio trains AI actors with consistent faces/styles
- Dzine AI Lip Sync 2.0 perfects video dubbing
- Google's Ironwood TPU delivers 42.5 exaflops with 2x efficiency
- ByteDance UNO enables multi-image subject transfer via FLUX
- NVIDIA launched Nemotron 253B with 4x faster AI inference
- Grok adding real-time search autocomplete feature soon
- MediaPipe Face Landmark tracks 468 facial points in real-time to animate 3D models instantly
- Ideogram now lets users pick 1-4 images per prompt
- OpenAI lets developers integrate Evals testing via new API
- Gemini 2.5 Pro enhances Deep Research with superior analytics

Thu, Apr 10
- Recraft adds Google's Imagen 3 for AI image creation
- OpenAI releases BrowseComp benchmark for AI web browsing skills
- Mira Murati's Thinking Machines Lab raises 2B at 10B valuation
- ChatGPT memory now uses full chat history for responses
- New Krea Stage tool builds full 3D scenes via AI
- Pika's new feature, Twists, seamlessly edits individual objects in videos
- Freepik adds seamless in-workspace video editing tool
- Unitree teases sci-fi robot combat livestream coming soon
- Alibaba unveils FantasyTalking, it animates photos with full-body talking videos
- n8n adds native MCP support with client/trigger nodes
- OpenAI prepares o3 and o4-mini ChatGPT model launches
- Midjourney V7 update includes image feedback to refine AI quality
- xAI plans Grok 3.5 and 4 releases for 2025
- Adobe developing AI agents for Photoshop and Premiere Pro
- FAL becomes Hugging Face Enterprise inference provider with billing API
- OpenAI countersues Musk alleging harassment and IP control attempts
- Grok-3 API launches with usage-based pricing tiers

Fri, Apr 11
- MecAgent brings Cursor-like AI tools to mechanical engineering
- Fal AI's FLUX.1 delivers previews in under 100ms
- LTX Studio introduces one-click upscale to 4K resolution
- Canva now AI-powered entire creative workflow from design to code
- Eleven Labs simplifies professional voice cloning with new PVC flow
- Build voice AI agents with ElevenLabs' new API guide
- RenderNet rebrands as Affogato, keeps AI media focus
- Google Veo-2 now free on AI Studio with video generation
- Pika Labs unveils hyper-realistic video object manipulation technology
- Higgsfield Mix enables limitless camera moves for filmmakers

Sat, Apr 12
- YouTube's AI-powered Music Assistant generates copyright-free tracks from text prompts
- Poe integrates Google Gemini 2.0 Flash native image generation