Have you heard these exciting AI news? - October 3, 2025 AI Updates Weekly

Have you heard these exciting AI news? - October 3, 2025 AI Updates Weekly Presented by Lev Selector Slides - https://github.com/lselector/seminar/tree/master/2025 (click on pptx file, then on "raw" or download button on the right) ------------------------------------------------------- Please write in comments if you have experience with both Claude Sonnet 4.5 and latest Gemini Pro for coding. Which one do you prefer? ------------------------------------------------------- Need AI Consulting ? - Enterprise AI Solutions - https://EAIS.ai - Linkedin - https://www.linkedin.com/in/levselector - GitHub - https://github.com/lselector ------------------------------------------------------- Our books: "Artificial Intelligence In Business: Strategies That Work" https://www.amazon.com/dp/B0FKZ3364G/ "AI Product Manager: Building Tomorrow" https://www.amazon.com/dp/B0FMS9NFK9/ --------- Contents of today's video: 00:00 AI updates arrive faster than your coffee cools 00:24 Crowd-sourced "LM Arena" Leaderboard 01:04 Anthropic's Claude Sonnet 4.5 01:37 GLM-4.6 Open Source King 02:53 Claude Code 2.0 03:38 OpenAI Sora 2 04:38 Perplexity Search API 06:11 ChatGPT Pulse for Pro Mobile Users 06:41 K2 Think 32B AI Reasoning Model by MBZUAI 07:20 OpenAI GDPval - measures model performance 07:38 Meta GAIA: a benchmark for General AI Assistants 08:04 Moonshot AI OK Computer 08:24 Gemini Robotics 1.5 with AI agents 08:42 Databricks commits to $100M in OpenAI spending 08:50 Distyl AI raises $175M in funding at $1.8B valuation 09:10 Alibaba Wan 2.2 Animate 09:40 Kling AI 2.5 Turbo 10:12 Qwen-Image-Edit-2509 & Qwen3-Max 10:54 xAI gives Grok 4 to US federal agencies 11:10 Meta FAIR - open-source Code World Model 11:23 Facebook adds an AI Dating Assistant 11:36 Emad Mostaque book "The Last Economy" 12:27 Abacus AI DeepAgent Desktop & CLI 13:00 Meta Vibes - AI Videos for Instagram 13:34 Anything - text-to-app startup 14:22 Tinker API for fine-tuning 14:58 Microsoft Unified Agent Framework 15:56 Cerebras Raises $1.1B at $8.1 valuation 16:13 Detecting hallucinations in RAG with 85% accuracy 16:55 Periodic Labs - AI Scientist 17:35 Genspark AI Workspace 19:31 IBM & HSBC - Quantum Computers to improve trading 19:46 Netflix: monolithic architectures can scale 20:26 StoneMQ - Open-source Messaging in Rust 21:12 RAG python modules 21:47 Traycer - AI coding assistant 22:13 Gemini 2.5 Flash Light model 22:55 Richard Sutton - Father of RL about LLMs 24:33 21 Things You Should Be Using AI For ... 25:29 Evaluate AI for Production 26:35 Rust instead of Python 28:34 Time to self-driving cars 30:16 Jobs, Layoffs, demand for AI Engineers *Takeaways:* 1. The pace of AI model development is accelerating Anthropic's Claude Sonnet 4.5 Zhipu AI's GLM-4.6 2. AI is increasingly being integrated into practice Perplexity Search API autonomous coding assistants like Claude Code 2.0 comprehensive business productivity platforms like Genspark AI Workspace. 3. Richard Sutton's critique of LLMs in favor of reinforcement learning 4. Deployment in physical world (for example Self-driving cars and trucks) is progressing much more slowly due to manufacturing and logistical hurdles *Quotes:* * "AI updates arrive faster than your coffee cools, which is kind of feels true." * "Ahmad [Mostaque] believes traditional economic structures will be obsolete within 3 years. He warns about the permanent underclass if AI benefits [are not] distributed properly." * "[Richard Sutton] believes LLM is a dead end and... LLMs mimic what people say not what actually happens in the world whereas reinforcement learning is about learning from direct experience and consequences." *Summary:* In this talk, I cover the weekly updates in artificial intelligence for October 3rd, 2025. I discuss the rapid release of new, highly capable models like Anthropic's Claude Sonnet 4.5 and Zhipu's GLM-4.6, which are excelling in coding and reasoning tasks. I also highlight the launch of powerful new tools and platforms, including OpenAI's Sora 2 for video generation, the highly accessible Perplexity Search API, and the move towards more autonomous agent frameworks. Finally, I explore broader industry trends, such as Richard Sutton's critique of current LLM technology, the growing adoption of Rust for AI development, and a realistic perspective on the deployment timeline for physical AI like self-driving vehicles. *Keywords:* Claude Sonnet 4.5, GLM-4.6, OpenAI Sora 2, Perplexity Search API, Claude Code 2.0, Reinforcement Learning, Richard Sutton, Emad Mostaque, Rust, Genspark AI, RAG (Retrieval-Augmented Generation), Self-driving cars, Cerebras, Microsoft Unified Agent Framework, Meta GAIA, Gemini Robotics, Zhipu.ai, Anthropic, StoneMQ, Agentic AI.