GPT-5 Pro generates a brand new mathematical proof

PLUS: Gemini Live will show you what it's talking about & Microsoft debuts VibeVoice AI. xAI sues Apple and OpenAI, Perplexity launches Comet+.

In today’s agenda:

1️⃣ OpenAI Researcher claims GPT-5-Pro developed a new mathematical demonstration improving on human knowledge.

2️⃣ Google's Gemini Live gains camera-based video interaction features

3️⃣ Microsoft releases VibeVoice-1.5B with 90-minute speech generation

  • xAI sues Apple and OpenAI over AI competition and App Store rankings

  • Perplexity launches Comet+ advanced search functionality

MAIN AI UPDATES / 26th August 2025

🧠 GPT-5 Pro generates a new mathematical proof 🧠
OpenAI's Researcher claims that GPT-5 generated a provably correct brand new mathematical demonstration.

OpenAI employee Sebastien Bubeck claims that GPT-5-Pro has moved beyond solving known probblems and has insteade developed a brand new mathematical proof. If counterproofed this could be the first time that an AI Model discovers new math, thus bringing the AI wave into “original contribution to human knowledge“ for the first time. When GPT-5 Pro was presented with a clean open problem in convex optimization, the model seems to have independently proved a better bound than what was originally published in the research paper provided. Although there is an ongoing discussion on the elegance of the new mathematical proof, it has been verified as mathematically correct by the online and research community, showcasing AI's growing potential in advancing mathematical research.

👁️ Gemini Live will show you what it's talking about 👁️
AI assistant will highlight part of the video stream coming from your camera to improve visual interaction.

Google's Gemini Live AI assistant is gaining visual interaction capabilities that will enable users to understand to which object or area of the video Gemini Live is referring to when giving guidance. The feature will initially roll out on the new Pixel 10 and other Android devices, enabling more intuitive interactions between users and the AI assistant. This represents a significant step toward multimodal AI experiences that combine voice, text, and visual understanding for more natural device interactions where Gemini can see and respond to what's happening in your video feed.

🤖 Microsoft debuts VibeVoice AI 🤖
A newly released opensource text-to-speech technology with multi-speaker capabilities

Microsoft has released VibeVoice-1.5B, an MIT-licensed open-source text-to-speech model that can generate up to 90 minutes of continuous speech with as many as 4 speakers. Built atop Alibaba's Qwen2.5 model, the tool features cross-lingual singing and powerful emotional control into ultra-realistic speech, making it ideal for AI-generated podcasts and dialogues. The model packs advanced capabilities while remaining completely open-source for developers.

INTERESTING TO KNOW

⚖️ xAI sues Apple & OpenAI ⚖️

Elon Musk's xAI has filed lawsuits against both Apple and OpenAI over AI competition and App Store ranking practices. The legal action alleges that these companies are engaging in anti-competitive behavior that undermines fair competition in the AI marketplace. The lawsuit specifically targets App Store rankings and AI competition practices, marking a significant escalation in the ongoing tensions between major AI companies as the industry continues to consolidate power among a few key players.

🔍 Perplexity launches Comet+ 🔍

Perplexity has introduced Comet+, an advanced search functionality that enhances the platform's AI-powered search capabilities. The new feature represents a significant upgrade to Perplexity's search infrastructure, offering more sophisticated query processing and result generation. Comet+ aims to provide users with more accurate and comprehensive search results through improved AI reasoning and knowledge synthesis.

📩 Have questions or feedback? Just reply to this email , we’d love to hear from you!

🔗 Stay connected: