
AI Toolz Podcast - 2026-10-19 : apertus drops radically transparent models
In today's episode: Apertus launches v1.5 of its open-weight models with native image and audio understanding, 262K context windows, and unprecedented transparency including full data provenance and training checkpoints. Also: Anthropic is privately pitching Congress a federal AI safety bill with an extremely high bar for intervention—$1B in damages or 50+ casualties. The company wants to preempt state regulations with a federal framework they help design.
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.