AI News · The field notes
The Daily AI Drop - Morning Edition, November 5, 2026
ChatGPT Images 2.6 with reference images and GIFs, Ripple-65B tested locally, a DIY PDF reader to replace Acrobat, and two takes on learning and careers in the age of AI.
Listen: Listen to this episode
Five new videos overnight: ChatGPT Images 2.6 brings reference images and GIFs to everyone, Ripple-65B shows open models running locally at serious scale, Matthew Berman rebuilds the PDF reader from scratch, and Aish Reganti and Aishwarya Srinivasan offer two honest answers to what you should actually learn now that AI can teach you anything.
Introducing ChatGPT Images 2.6 — OpenAI
OpenAI's image model inside ChatGPT now makes GIFs and supports reference images, so a character or style stays consistent across generations. This is the consumer-side follow-up to the faster API generation news from yesterday evening — the version millions of people will actually open on their phones. If you make slides, social posts, or product mockups, the reference image feature is worth trying this week.
Ripple-65B: Thin and Deep Dense Model, Tested Locally — Fahd Mirza
Fahd installs and tests Ricursive Intelligence's Ripple-65B Preview, a 65-billion-parameter dense Transformer. A model of this size running on local hardware is a statement about where open models are headed — serious capability without a data center. Fahd's real local tests are worth more than a benchmark chart if you're curious about running models yourself.
I built my own PDF reader to replace Adobe Acrobat — Matthew Berman
Matthew builds his own PDF reader as an alternative to Acrobat, one of those tools everybody pays for and nobody loves. The bigger point: the software we treat as unchangeable is increasingly rebuildable with modern AI tooling. If you read research papers or reports all day, a reader tuned to your own habits is genuinely appealing.
Did AI make my $100K CMU master's useless? — Aish Reganti
Aish did her master's at Carnegie Mellon, the top-ranked graduate computer science program, and says she would still go — for a different reason than most people expect. Drawing on Randy Pausch's idea of "head fakes," she argues the degree was never really about the lectures, but about the people, the pressure, and the environment. A useful reframing if you're weighing a degree, a bootcamp, or learning on your own.
14 Day Roadmap to learn AI Agents — Aishwarya Srinivasan
A focused two-week, step-by-step plan for learning AI agents, from the basics to building something real. Agents are the skill everybody is hiring for and nobody quite knows how to teach yet — a structured fourteen-day plan beats another month of bookmarked tutorials you never open.