Today in AI

OpenAI introduced GPT-5.6 not as a single model but as a three-tier lineup: Sol, positioned as the flagship workhorse; Terra, a mid-tier option; and Luna, a budget-focused release. CEO Sam Altman said Sol is roughly 54% more token-efficient on coding tasks than its predecessor. Splitting the release into tiers instead of one model is a clearer acknowledgment that price and speed now matter as much to customers as raw capability.

Anthropic launched "Reflect," a dashboard inside Claude that lets users track and visualize their usage habits, alongside tools to set quiet hours or schedule reminders to step away. The framing is wellness-oriented, but the feature also doubles as a way to surface more use cases back to the user and encourage habitual use. It's a small but telling example of AI companies borrowing directly from social media's engagement playbook.

Gradium, a Paris-based startup building voice AI models, raised $100 million total across its seed round, reopened to new investors including Nvidia. The size of the round is unusual for a seed stage — a sign of how much capital is still chasing voice AI specifically, even as the broader funding market has gotten choosier. European AI startups landing checks this large from U.S. and Nvidia-adjacent investors is also worth watching as its own trend.

The New York Times and the Daily News allege OpenAI misrepresented its ability to search its own chat logs and training data for their copyrighted material, escalating a two-year-old lawsuit over whether ChatGPT was trained on and reproduces Times journalism. The claim, if substantiated, could affect how much of OpenAI's internal data practices get exposed during discovery. It lands the same week Anthropic's own copyright case reached a $1.5 billion resolution, raising the stakes on how this one gets settled.