Cassidy Hardin from Google DeepMind introduces Gemma 4, a new family of open-weight models with significant architectural and performance improvements. This summary covers the four new models (31B Dense, 26B MoE, and two "Effective" on-device models), deep dives into architectural changes like mixed global/local attention and Per-Layer Embeddings (PLE), and details the new native multimodal capabilities for vision and audio.
A deep dive into Google DeepMind's Gemma 4, the latest family of open models. This summary covers the new model architectures like per-layer embeddings, on-device agentic capabilities, multimodal features, and the growing ecosystem of fine-tuned applications from medicine to sovereign AI.
IBM's Granite 4.0 models introduce a groundbreaking hybrid architecture combining Mamba-2 and Transformer blocks with a Mixture of Experts (MoE) design. This approach enables smaller models to achieve superior performance, speed, and memory efficiency, even outperforming much larger models on key enterprise tasks while running on consumer-grade hardware.
A deep dive into the latest AI model releases, including IBM's hyper-efficient Granite 4.0, Anthropic's code-focused Claude 4.5, and OpenAI's consumer-centric Sora 2. The discussion covers the strategic differentiation between major AI labs, the future of open-source, the rise of AI e-commerce agents, and the emerging cybersecurity challenges of social engineering AI.
Andrew Rabinovich, CTO and Head of AI at Upwork, details their strategy for building AI agents for digital work. He introduces a custom reinforcement learning approach called RLEF (Reinforcement Learning from Experience), explains why digital work marketplaces are ideal training grounds, and shares his vision for a future where AI delivers finished projects, orchestrated by a meta-agent named Uma.
A deep dive into seven essential AI concepts shaping the future of intelligent systems, including Agentic AI, RAG, Mixture of Experts (MoE), and the theoretical frontier of Artificial Superintelligence (ASI).