In this video I break down three signals that tell you where the AI market is actually heading: a leaked Anthropic product called Claude Operon, a stealth Gemma 4 sighting on Chatbot Arena, and the reality check that is ARC-AGI-3. First up, Claude Operon looks like a brand-new standalone mode inside the Claude desktop app—sitting next to Chat, Code, and Co-Work—but built specifically for scientific research in biology and health. The onboarding and suggested tasks are extremely targeted: phylogenetic trees, CRISPR knockout screens, single-cell RNA-seq analysis, and ranking enzyme variants with protein language models. This is Anthropic moving past “chatbot” and into a dedicated research workspace with plan/auto modes, project-level system prompts, artifacts, and local file access—exactly what real labs need when they’re dealing with huge datasets on institutional machines. Then I cover Gemma 4 being tested on Arena under a codename (“significant otter”), plus the Hugging Face collection update pattern that usually precedes a release. The leaked card suggests multiple sizes (2B and 4B) with multimodal support, which matters a lot for local-first devs and edge deployments. Finally, ARC-AGI-3 dropped and basically humbled every frontier model—sub-0.5% scores across the board—reminding us of the point I keep coming back to: we are genuinely early, and real generalization is still largely unsolved.