ThursdAI - The top AI news from the past week

Wolfram Ravenwolf, an AI engineer known for technical commentary on model releases, joins to unpack this whirlwind week. They cover Genie 3’s real-time persistent worlds and Chrome’s new agentic browsing for billions. Open-source breakthroughs like Kimi K2.5, Trinity Large, Grok Imagine’s API and advances in video and image models get technical analysis and reactions.

Dan Pagan, a tech entrepreneur and early adopter of CloudBot, joins to explore the innovative personal AI assistant that can improve itself and manage tasks through phone control. He shares insights on CloudBot's architecture, skills setup, and local security implications. The discussion highlights its impressive capabilities like code writing and life organization. Additionally, the team delves into GLM 4.7 Flash's advancements and Anthropic's newly released constitution, which sets ethical standards for AI behavior.

Eleanor Berger, founder of Agentic Ventures and an expert in agent skills, dives into the exciting world of skill-driven AI tools. She explains how to customize AI agents simply and effectively, ensuring they stay contextually relevant. Discover the fascinating capabilities of Anthropic's Claude Cowork for non-coders and the latest from OpenAI's GPT-5.2 Codex. Eleanor also discusses building reusable skills that can replace apps, showcasing the potential for shared expertise in AI development. Get ready for a personalized AI future!

Join Ryan Carson, a product innovator and founder, who recently gained viral fame for his unique take on coding with Ralph Wiggum! He dives into the fascinating world of agentic coding techniques. The discussion also highlights NVIDIA's groundbreaking Vera Rubin platform, boasting 5x improvements. Excitement mounts around AI advancements in healthcare with the launch of GPT Health. Plus, the panel explores innovative open-source models like Solar Open 100B, emphasizing efficiency and agent performance. Tune in for insights that bridge humor and cutting-edge tech!

Will Brown, a researcher at Prime Intellect and creator of the Verifiers library, shares his insights on reinforcement learning environments and his humorous takes on AI discussions online. He highlights the importance of structured formats in model training and discusses scaling laws and their implications on cost-quality tradeoffs. Will also delves into the concept of optimizing thinking length for AI models, and how verification functions can improve outputs. He explores unique examples from the AI Engineer conference and the value of community in advancing AI research.

Nisten Tahiraj, a frequent contributor focused on open-source models, joins the discussion to recap a landmark year in AI. They dive into the explosive rise of agent-based models and reasoning, spotlighting significant releases like DeepSeek R1 and Cloud Code. Nisten elaborates on VibeCoding's impact on developers and the evolution of multimodal reasoning models. The conversation also highlights the fierce competition among AI labs and the provocative developments in real-time media and audio technology.

This week features Wolfram Ravenwolf, an AI evaluator focused on benchmarking, who discusses performance metrics and his new role at Weights & Biases. Jan Peleg, an AI practitioner, dives into the technicalities of multimodal models and the impact of open-source releases. Quindla Haltman-Kramer, a voice/audio expert, shares insights on Grok Voice, its integration with Tesla, and audio robotics. The panel also covers the explosive arrival of Gemini 3 Flash and OpenAI's GPT 5.2 Codex, highlighting groundbreaking innovations in AI.

Wolfram Ravenwolf, a frequent panel contributor known for his deep dives into model behavior, and Jan Peleg, an insightful commentator on model performance, share their thoughts on the groundbreaking launch of GPT-5.2. They discuss its advancements in long-horizon reasoning and reduced hallucinations. The conversation takes an exhilarating turn as they explore the implications of AI in space with StarCloud’s H100 satellite. The episode wraps up with discussions on open-source innovations and new integrations that are shaping the AI landscape.

Lucas Atkins, CTO of RCAI and leader in U.S.-based MOE models, dives into the significant launch of Trinity models and their enterprise implications. He highlights the importance of training and compliance in model development, explaining the efficiency of MOE inference and the challenges of scaling. The conversation shifts to the competitive benchmarks of DeepSeek V3.2, which showcases exceptional performance. Exciting insights on the latest AI integrations wrap up the discussion, emphasizing real-world applications and the rapid evolution of AI technology.

In this engaging discussion, Kevin Hou, the Product Engineering Lead at Google DeepMind, dives into Antigravity, Google's new Agentic IDE, highlighting its unique features like artifacts and agent managers. Ammaar Reshi and Kat Kampf, leaders at Google AI Studio, explore the exciting AI Studio experience and the launch of Gemini 3, discussing multimodality and user-friendly integrations. They also share insights into building apps with Nano Banana Pro and the evolution of generative UIs, showcasing the transformative potential of AI in app development.