Last Week It Quietly Charged Me More

Last Wednesday afternoon I was using Claude to polish a collaboration proposal — same as any other day — when it cut me off mid-draft: "Daily usage limit reached." I hadn't changed my plan. I hadn 't changed my habits. I just had this nagging feeling: why am I burning through quota so much faster than last month? Turns out Anthropic had quietly rolled out a new version, and the rules changed with it.

If you rely on AI every day for writing, organizing information, or building client proposals, you know exactly that feeling — "the bill didn 't change, but somehow I'm getting less."

What Actually Happened — In Plain Language

Anthropic released Claude Opus 4.7 (think of it as swapping in a smarter brain). Officially , "pricing is the same as the previous version" — but two things changed underneath:

  • Hidden change #1: the way characters are counted. AI charges by " tokens" — basically the unit for measuring text. In the new version, the same passage registers as more tokens than before, so your quota drains faster.
  • Hidden change #2: "Extended Thinking" is on by default. Before answering you, the AI now runs through a hidden reasoning pass in the background. The output quality is better — but it burns several times more quota than normal mode , and it's switched on automatically, not by your choice.

Lena, a freelance copywriter in Guangzhou, told me she opened three client copy tasks on Monday morning and hit her limit by 2 p .m. — the exact same workload had lasted her until evening the month before. She thought she'd just been using it too much. She hadn't. The version had quietly upgraded.

What You Can Do Today — Costs and Steps

Money: Nothing extra. Your existing account is fine.
Time: Five minutes once you finish reading this.
Technical barrier: If you can click a button on a webpage, you're good.
First step: Open Claude's chat interface and click "Model" in the top-right corner to see which version you're currently on.

From there, two things worth doing:

  1. Manually turn off Extended Thinking. In the bottom-left of the chat box, look for a small icon (a lightbulb or gear). If you see a toggle labeled "Extended Thinking" or similar, switch it off. For everyday copywriting and email edits, normal mode is completely sufficient.
  2. Get into the habit of short, focused conversations. Try to keep each conversation to one task. Don't go back and forth rev ising something fifteen times in a single thread — every exchange burns quota. Breaking a big task into several separate conversations actually costs less than trying to get it all done in one go.

I got this wrong myself — I assumed iterating inside one conversation was more efficient. It was actually doubling my usage.

Do You Actually Need to Deal With This Right Now?

Not everyone needs to act immediately . Here's how I'd break it down:

If you're just starting out with AI and using it less than 30 minutes a day: Don't worry about it for now. You're nowhere near the limit. Come back to this when you actually hit "quota exhausted" for the first time.

If you already have one or two steady clients and rely on AI for real daily work: I'd suggest spending five minutes this week to turn off Extended Thinking as described above. It's not urgent — but it'll stretch your quota by a few extra days each month .

If you're scaling up, have team members using it, or are running several projects in parallel: It's worth auditing your daily use cases seriously — which tasks genuinely benefit from Extended Thinking (complex analysis, for example) and which absolutely don't (changing a headline). Using them selectively can meaningfully increase what you get out of the same budget.

Bottom line: the new Claude is genuinely better. There's just one small trap to know about. Now that you know, step around it and keep going .