OpenAI's new flagship ships in three sizes with aggressive pricing — Luna at $1/$6, Terra at $2.50/$15, Sol at $5/$30 per million tokens. That undercuts Claude Opus and positions OpenAI to dominate agentic workflows where token costs compound fast.
Patreon partnered with Cloudflare to wall off creator content from AI training scrapers. CEO Jack Conte's message was blunt: consent or stay out. This is the clearest platform-level stance on creator data rights yet.
Real browsing data shows AI-generated content now saturates both platforms at levels most users would find alarming. The feeds aren't just noisy — they're increasingly synthetic.
New research confirms AI-generated stories are easy to spot — ChatGPT leans on dream sequences, Gemini obsesses over character description. The tells are consistent enough to be almost funny.
Meta's first Spark model to offer API access claims meaningful gains in agentic tool calling and computer use. With GPT-5.6 and Grok 4.5 also shipping this week, the frontier model market is in full sprint.
Meta's Chief Data Officer is publicly framing AI agents as the successor architecture to traditional e-commerce. When a company with Meta's distribution says it out loud, the roadmap is already set.
Apple has sued OpenAI for alleged misappropriation of trade secrets. This is the most significant legal collision between big tech incumbents and AI labs to date — and signals that the era of quiet tension between Silicon Valley's old guard and the new AI layer is officially over.
Sebastian Raschka dropped a clean tactical note: for agentic coding, a Luna model on high effort beats Terra on standard — same or better results at lower cost. As the GPT-5.6 family lands, practitioners are already reverse-engineering the optimal cost-performance stack.
The AI 2027 authors are back, this time not predicting but proposing — their new AI 2040: Plan A report lays out an optimistic trajectory. Worth watching: the people closest to thinking about AI risk are now pivoting to writing the future they'd prefer to see.
Ethan Mollick is already joking about artisanal model selection — Luna xHigh for recipes, Sol Medium for sonnets — but the joke contains the real insight: we're entering a world of model sommelier skills, where knowing which model to pick for which task becomes a genuine professional edge.