Save This

5 Things Anthropic
Changed In
Claude 4.7

Claude Opus 4.7 prompts differently than 4.6. If your outputs have felt off since the upgrade, here’s exactly why — pulled from Anthropic’s official prompting guide, with the exact phrasings to use.

Why Your Outputs Got Weirder

If you haven’t updated how you prompt since Claude Opus 4.7 dropped, you might have noticed the outputs getting shorter, more literal, or weirdly off. That’s not a regression — it’s a deliberate set of behavior changes Anthropic published. The fix is a handful of small phrasing updates. Here they are, with the verbatim quotes from the official guide.

Change 1 4.7 Interprets You More Literally

In 4.6, “clean up this draft” would fix typos, tighten sentences, and restructure the whole thing. In 4.7 it’ll do exactly what you said and nothing more. Anthropic puts it this way:

“Claude Opus 4.7 interprets prompts more literally and explicitly than Claude Opus 4.6, particularly at lower effort levels. It will not silently generalize an instruction from one item to another, and it will not infer requests you didn’t make.” — Anthropic prompting guide, “More literal instruction following”

The fix: spell out every step. Replace vague verbs with explicit lists.

Old (4.6) — worked fine

“Clean up this draft.”

New (4.7) — works much better

“Fix any typos, tighten any sentence over 25 words, and remove anything unnecessary. Keep the structure.”

Change 2 Response Length Now Auto-Adjusts

4.7 calibrates length to how complex the task looks. “Summarize this” gets you one sentence. “Walk me through this” gets you several paragraphs. Anthropic:

“Claude Opus 4.7 calibrates response length to how complex it judges the task to be, rather than defaulting to a fixed verbosity. This usually means shorter answers on simple lookups and much longer ones on open-ended analysis.” — Anthropic prompting guide, “Response length and verbosity”

The fix: if you want depth, explicitly ask for depth.

Gets you 1 sentence

“Summarize this.”

Gets you a real breakdown

“Give me a full breakdown of this in 3-4 paragraphs, not a summary. Cover the main argument, the evidence, and the counterargument.”

Change 3 Stop Using ALL CAPS and “You MUST”

If you developed habits like writing “CRITICAL:” or “YOU MUST DO X” in your prompts to push older models, drop them — 4.7 overcorrects on aggressive phrasing and starts overtriggering. Anthropic:

“If your prompts were designed to reduce undertriggering on tools or skills, these models may now overtrigger. The fix is to dial back any aggressive language. Where you might have said ‘CRITICAL: You MUST use this tool when…’, you can use more normal prompting like ‘Use this tool when…’.” — Anthropic prompting guide, “Tool usage”

The fix: just talk to it normally.

Old habit — now overtriggers

“CRITICAL: You MUST use the calculator tool for ANY math.”

Normal phrasing

“Use the calculator for math problems.”

Change 4 Add “Go Beyond the Basics” for Creative Work

For anything creative — design, writing, building from scratch — 4.7 will give you exactly what you asked for and stop. There’s a magic phrase from the Anthropic guide that pushes it past the bare minimum:

“Create an analytics dashboard. Include as many relevant features and interactions as possible. Go beyond the basics to create a fully-featured implementation.— Anthropic prompting guide, “Be clear and direct” (their example, emphasis ours)

The fix: add “go beyond the basics” to any creative prompt.

Gets you a minimum-viable version

“Design a landing page for my app.”

Gets you something to actually ship

“Design a landing page for my app. Go beyond the basics — include hero, social proof, feature breakdown, pricing, FAQ, and a final CTA.”

Change 5 4.7 Doesn’t Reason by Default — Turn It On

Extended thinking is OFF by default in 4.7 unless you explicitly enable it. That’s a big difference from how earlier reasoning-capable models behaved. Two ways to fix it:

If you’re using the API: set the effort parameter to high or xhigh for tasks that need multi-step reasoning.
If you’re prompting in chat: add Anthropic’s actual recommended phrasing.

“This task involves multi-step reasoning. Think carefully through the problem before responding.” — Anthropic prompting guide, recommended phrasing

The fix: for any prompt that needs real thought, add the line above. I’ve A/B tested it on the same tasks and the difference in output quality is consistent.

Save This The 4.7 Prompting Cheat Sheet

Bookmark this. Use it the next time your Claude output feels weirdly short, off, or stuck at the bare minimum.

If it’s being too literal: spell out every step explicitly.
If the answer is too short: ask for “a full breakdown, not a summary.”
If you wrote “CRITICAL” or “MUST”: rewrite normally.
If you want it to push: add “go beyond the basics.”
If it needs to think: add “Think carefully through the problem before responding,” or set effort to high/xhigh in the API.

The Real Win

None of these are workarounds — they’re the operating manual. Anthropic published this for a reason: 4.7 isn’t worse than 4.6, it just expects you to be more explicit about what you want. The 90 seconds of phrasing change is the entire upgrade.

The Only AI Masterclass You Need

Build AI Systems That Run Your Work, Business, And Life

If this guide helped, but you’re looking to go deeper, I got you!! My 30-Day Challenge takes you from saving AI tips you never use to actually building with AI, step-by-step.

I show you exactly how I automated two e-commerce brands, my social media, and most of my personal life, then hand you the agents, workflows & systems to do the same. I’m teaching you every single thing I know with one lesson and one build a day.

Join the AI Masterclass →

© 2026 Mariah Brunner. All rights reserved.