Plus: rewrite any document in plain, short sentences with your AI chat in four steps.

From our library · 2026-11-11
Anthropic's filing for its planned stock listing shows about 85.5% gross margin for the first nine months of 2026. That figure counts only the cost of running Claude for paying customers: $33.2 billion in revenue against $4.812 billion in cost of revenue. Training new models and free-tier use sit in other lines. Count all computing costs together, and about 43.4% of revenue is left, by our own arithmetic. The company also reported a $1.3 billion operating loss under standard accounting.
Why it matters: A high margin on serving Claude does not mean the whole business is highly profitable. When you read a margin headline, ask which costs it leaves out.
What you can do: Next time you see a profit figure for an AI company, check which costs it counts. Compare it with the loss or the total spend.
Read the sourceNews · 2026-11-16
DeepSeek reportedly limited its free web and app chat on November 16. Users describe a daily conversation quota, a roughly 40,000-character cap per message, and deletion of long chats idle for a week. DeepSeek's notices cite too little computing power.
Why it matters: Free AI chat is not unlimited. If you rely on it for work, save important chats and consider a paid, metered option.
Read the sourceNews · 2026-11-16
Tencent released Hy4 flash on November 16, a cheaper model for everyday office work such as search, documents and slides. Tencent says its average task cost is about one-twentieth of full Hy4 in its own tests. It scores lower on hard coding tests.
Why it matters: Tencent points routine work to the cheaper tier and hard coding to full Hy4. The cost figures are Tencent's own, not independent.
Read the source
News · 2026-11-16
miHoYo's research team released Gamer, an AI model that plays long 3D game quests by reading the screen and sending keyboard and mouse input. It does not read hidden game data. It is research code, not a product players can use today.
Why it matters: Games are a hard test for AI that must plan over an hour. The reported results come from the team's own tests.
Read the sourceFrom our library · 2026-10-14
ASD-STE100 is a plain-writing standard built for maintenance manuals. A short instruction can make an AI chat tool write clear, short sentences for setup guides, error messages and reports. It is a poor fit for marketing or blog writing.
What you can do: 1. Pick one document with a clear reader, such as a setup guide. 2. Paste it into your AI chat and write: Rewrite this in ASD-STE100. Do not add facts. Do not remove steps. 3. Compare the result with your original and check that steps, numbers and warnings are intact. 4. Ask one person to follow the steps without your help.
Read the source
From us
Two live evenings for professionals who want Claude to do more than answer one-off questions. No coding needed. You build a Claude Project for your role, an agent skill, and a personal prompt library to take home.
See the details
From us
Two live evenings for people who work in Microsoft 365 every day. You practise Copilot inside Excel and other Microsoft apps, see Copilot Cowork handle multi-step tasks, and build an agent and an automation in Copilot Studio. No coding needed.
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…