Plus: give Claude a whole task with a clear finish line in four steps

News · 2026-11-17
OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on November 17. Astra is built for long professional work like slides, spreadsheets and legal documents, and OpenAI cut the price developers pay for it by 20%. Sol is the everyday model and keeps its old price. Paid ChatGPT plans and Codex get both the same day.
Why it matters: Cheaper long-work models make AI easier to afford for office tasks, but the benchmark wins are OpenAI's own claims. Test them on your own work.
What you can do: If you pay for ChatGPT Plus or Pro, try GPT-6.2 Sol on one real task this week. Compare the result with your usual model before you change habits.
Read the sourceNews · 2026-11-17
At its Ignite event on November 17, Microsoft released seven in-house MAI models. The lead one, MAI Code 2, helps with everyday coding in GitHub Copilot. Microsoft claims it reaches 98% of GPT-6.1 Sol's coding quality at about 40% of the price.
Why it matters: Companies get another coding helper to compare. These numbers are Microsoft's own, from four tests it picked, so treat them as a starting point.
Read the source
News · 2026-11-17
Security firm Calif disclosed Java4Shell, a critical flaw in how Java decodes text on some server chips. Oracle released fixed versions on November 17. Calif says AI helped find it, and the New York Times reports OpenAI's GPT-6 Cyber was involved.
Why it matters: Java runs many company servers, so IT teams should update and restart them. At disclosure, no attacks in the wild were confirmed.
Read the sourceNews · 2026-11-16
Reuters, citing the Wall Street Journal, says the administration will name more than two dozen US AI firms as national champions, including Nvidia, Amazon, Microsoft, OpenAI and Anthropic. Reuters could not verify it. A government site already lists priority export licensing and financing help.
Why it matters: The aim is to help US firms sell abroad. Nothing changes for chip rules yet, because the list is reported, not published.
Read the sourceFrom our library · 2026-09-23
Anthropic's own guide for Claude Opus 5.5 says to hand over the whole task in one message. Say what done looks like and when Claude should stop and ask you. It also says to drop lines like think carefully.
What you can do: 1. Write your whole task in one message. 2. Add one line that says what a finished result looks like. 3. Say when Claude should stop and ask you a question. 4. Delete any think carefully lines from your saved instructions.
Read the source
From us
Two live evenings for beginners. You build a knowledge base, an agent skill, a prompt library and your own Artifacts, then connect Claude to the apps you use. It suits non-technical professionals who want to become the AI person on their team.
See the details
From us
Two live evenings for beginner to intermediate learners. You explore Microsoft 365 Copilot, use it inside Excel and other Microsoft apps, put Copilot Cowork to work and build automations in Copilot Studio. The early-bird price is $103 until December 12.
See the detailsOne vote per browser. An informal reader poll, not a representative survey.
Comments
Join in with your explainx community account. Comments are plain text (no links, up to 600 characters), follow the community guidelines, and are reviewed before they appear.
No comments yet. What did you think of this edition?
Loading…