- First LLM trained in space
- Gemini now built into Apple iPhones and iPads
- Disney invests $1B into AI
A collection of AI Updates! 🧵
1. OpenAI Rolls Out GPT-5.2 to Everyone
GPT-5.2 Instant, Thinking, and Pro are live for Free/Plus/Pro/Business/Enterprise. Also in API and Codex. First model hitting human expert level on 44-occupation knowledge work eval including spreadsheets and presentations.
State-of-the-art long-context reasoning.
2. Google Rolls Out Built-In Gemini AI on Apple Devices
Gemini AI experience now available on iPhone and iPad. Puts Google's AI stack directly inside Apple's widely popular technology.
Major new channel of reach for Gemini.
3. Disney Invests $1B in OpenAI - Characters are Coming to Sora
Users can generate videos and images with Disney characters including Mickey Mouse, Ariel, and Cinderella across Sora and ChatGPT Image. Covers Disney, Marvel, Pixar, and Star Wars IPs.
Huge entertainment IPs are now available for AI generation.
4. AI Trained in Space for the First Time
Starcloud-1 used Nvidia H100 to train nano-GPT model on Shakespeare's complete works and ran inference successfully. Also ran inference on Gemma model.
Goal: move computing off Earth to leverage abundant solar energy.
5. xAI Launches World's First Nationwide AI Tutor Program
Partnership with El Salvador brings personalized Grok tutoring to 1M+ public-school students across 5,000+ schools over next two years. Empowers thousands of teachers as collaborative partners.
Will other LLMs try partnerships like this?
6. Microsoft AI Accelerates Cancer Discovery
GigaTIME simulates spatial proteomics from pathology slides, enabling population-scale tumor microenvironment analysis across dozens of cancer types. Partnership with Providence and UW. Published in Cell today.
AI revealing links between mutations, immunity, and outcomes.
7. Google Labs Launches Disco - Remix Open Tabs Into Custom Apps
GenTabs uses Gemini 3 to understand complex tasks through open tabs and chat history, then creates interactive web apps. Describe the tool you need and refine with natural language. Links back to original sources.
Aims to turn browser chaos into functional apps.
8. Google Releases Updated Gemini 2.5 Flash Native Audio
Now available via Live API with sharper function calling, robust instruction following, and smoother conversations. Also launching TTS updates with emotional style versatility, context-aware pacing, and improved multi-speaker capabilities.
Major voice and audio improvements.
9. Grok Voice Mode Gets Major Upgrades
Point camera at anything and ask "What am I looking at?" for instant analysis. Scans notes, places, paintings, documents, translates languages. Real-time web search.
Talk to Grok like a normal person in your language.
10. OpenAI Realtime API Gets Major Audio Improvements
89% fewer hallucinations in transcription, 35% fewer word errors in TTS, 22% better instruction following, 13% better function calling. Stronger in Chinese, Japanese, Indonesian, Hindi, Bengali, Italian.
Audio reliability significantly upgraded.
That's a wrap on this week's AI News.
Which update are you trying first?
LMK what else you want to see | More weekly AI + Agentic content releasing ever week!
[link] [comments]