Claude Opus 5.5 prompting tips are official guidance Anthropic published on 12 November 2026 for its default Claude model. Anthropic says to drop “think carefully” lines, start at Medium effort, describe the finished result in one message and attach charts as images. Opus 5.5 is the default on paid Claude plans.
Many Indian shop owners, coaches and freelancers still use prompts they wrote a year ago for older chatbots. Anthropic now says some of those habits slow Claude down without making answers better. Updating a few lines can save time and plan limits, which matters when you pay around ₹2,000 a month for Claude Pro. If you are new to AI at work, our AI for small business in India guide covers the basics first.
Key takeaways
- Anthropic published Claude Opus 5.5 prompting tips on 12 November 2026 for Claude app users.
- Opus 5.5 always reasons on its own, so “think step by step” lines are mostly unnecessary.
- Effort (Low to Max) is now the main control; Anthropic recommends starting at Medium.
- Ask for the finished file, attach charts as images and give the full brief in one message.
- Claude Pro costs about ₹2,000 a month in India, according to Indian news reports.
At a glance
| Product | Made by | Launched | Price | Available in India |
|---|---|---|---|---|
| Claude Opus 5.5 (prompting guidance) | Anthropic | Guide published 12 November 2026 | Paid Claude plans from about ₹2,000/month; API $4 / $20 per million input / output tokens | Yes, Claude is available in India; Opus 5.5 on the free plan is not confirmed |
What is Claude Opus 5.5?
Claude Opus 5.5 is a large language model made by Anthropic, the US company behind the Claude assistant. According to Anthropic, it is now the default model on the Pro, Max, Ultra and Team plans in the Claude app.
Anthropic describes it as its recommended “daily driver” for research, analysis and finished documents, decks and spreadsheets. Its bigger sibling, Claude Fable 5.2, is meant for the hardest long-running reasoning and coding work.
The news here is not a brand-new model. It is a set of official prompting tips. Anthropic published an article for Claude app users on 12 November 2026 and also keeps a longer technical guide, “Prompting Claude Opus 5.5”, in its developer documentation. The Decoder reported on the guidance on 14 November 2026.
The core message is simple. Opus 5.5 always reasons on its own, so many tricks people used to push older models to think harder are no longer needed.
Key features
- Thinking is always on: Anthropic says the model decides how much to reason based on the task and the effort level you pick. Lines like “think step by step” or “think carefully” are mostly unnecessary.
- Effort setting as the main control: You choose Low, Medium, High, Extra high or Max effort from the model menu next to the send button. Medium is the default.
- Medium matches the older High: In Anthropic’s own testing, Opus 5.5 at Medium matched or beat Claude Opus 5 at High on its evaluations.
- Better chart and image reading: Anthropic says even at the lowest effort, Opus 5.5 reads values off dense charts more accurately than Opus 5 did at its highest.
- Faster output: The developer guide says Opus 5.5 generates output tokens more than 30 percent faster than Opus 5 and often finishes the same task with fewer tokens.
- Fewer wrong figures: Anthropic says the model is much less likely to state an incorrect number or cite the wrong source in knowledge work.
- Stronger defence against hidden instructions: According to Anthropic, it resists instructions hidden inside web pages or pasted emails better than any earlier Opus model.
What Claude Opus 5.5 prompting tips does Anthropic give?
In short: write less “think harder” text and more clear detail about the result you want. Here is a plain summary of what Anthropic recommends, based on its official article and developer guide.
1. Remove “think hard” lines
If your saved prompts or custom instructions tell Claude to “think carefully before answering”, Anthropic suggests removing them. In its testing in a chat product, removing such a line made replies start sooner with no clear drop in quality.
2. Which Claude effort setting should you start with?
Anthropic says Medium, the default. Do not jump to Max for every task. Anthropic says Extra high and Max cost more and do not always improve results. Move to High only when Medium gets stuck. Before switching to a bigger model, try a higher effort first.
3. Describe the finished result in one message
Instead of giving instructions in small pieces, explain the goal, the output format and when Claude should ask you questions, all in one go. Anthropic says Opus 5.5 can often go from request to finished document in one pass.
4. Ask for the actual file
If you need a spreadsheet, a slide deck or a report, ask for that file directly instead of asking for text you then copy into Excel or PowerPoint. We covered Claude’s built-in file features in our post on Claude Sheets and Whiteboards for Indian users.
5. Attach charts and tables as images
You no longer need to type out numbers from a chart or a photo of a bill. Attach the image or file. Anthropic says image reading improved a lot in this model.
6. Name the design styles you do not want
Saying “avoid a generic AI look” mostly swaps one default style for another, according to Anthropic. It works better to name specific things, such as a cream background or pill-shaped buttons.
7. Share your house style and connect your files
Give Claude a sample of your brand tone, past documents or templates. Anthropic also recommends using connectors so the model can pull the information it needs from your tools instead of you pasting it each time.
Price and availability in India
Is Claude Opus 5.5 available in India?
Yes. Claude is available in India, and Anthropic started showing rupee prices for its plans in July 2026. The figures below come from Indian news reports and Anthropic’s pricing pages; plan prices include GST according to those reports. Always check claude.com/pricing before you pay, as prices can change.
| Plan | Price | Best for |
|---|---|---|
| Free | ₹0 | Trying Claude; whether Opus 5.5 is included on Free is not confirmed |
| Pro | About ₹2,000 per month (reports vary from ₹1,999 to ₹2,033) | Freelancers, students, small business owners |
| Max 5x | ₹11,999 per month | Heavy daily users and consultants |
| Max 20x | ₹23,999 per month | Power users who hit limits often |
| Team / Ultra | Rupee price not confirmed here | Agencies and companies with several users |
| API (developers) | $4 per million input tokens, $20 per million output tokens (roughly ₹350 and ₹1,760, approximate) | Apps, chatbots, automation |
Rupee values for the API are approximate and depend on the exchange rate. Anthropic also lists cheaper models: Sonnet 5.5 at $2 and $10, and Haiku 5.5 at $0.10 and $0.50 per million input and output tokens.
What would the API cost a small Indian business?
Here is a rough Cracckk estimate, not a test result. Say a shop’s chatbot answers 1,000 customer questions a month, with about 500 input tokens and 300 output tokens each.
That is 0.5 million input tokens (about $2) and 0.3 million output tokens (about $6) on Opus 5.5, so roughly $8 or about ₹700 a month, approximate. The same job on Haiku 5.5 would cost well under ₹50. Real costs depend on prompt length, effort level and the exchange rate.
Which Claude model should you use?
Anthropic’s own advice gives a simple ladder. It may help you save plan limits.
| Model | Anthropic’s suggested use |
|---|---|
| Haiku 5.5 | Quick lookups and high-volume short tasks |
| Sonnet 5.5 | Clear, well-scoped jobs, such as turning finished notes into a document |
| Opus 5.5 | Daily research, analysis and finished documents that need judgment |
| Fable 5.2 | The hardest, long-running reasoning and coding work |
Anthropic also says that if you regularly push Sonnet to Extra high or Max, its cost per task comes close to Opus 5.5, so switching to Opus may make more sense. Higher-priced models use up plan limits faster.
How Indian businesses can use it
D2C store owners
A small clothing or skincare brand can upload a screenshot of its monthly sales chart and ask for a one-page summary with three actions for next month. Since Opus 5.5 reads charts better, there is no need to type numbers in manually.
Describe the full output once: who will read it, the format, and the tone. That follows Anthropic’s “finished result in one message” tip.
Coaching centres
A coaching centre in Kota or Pune can ask for a full weekly test schedule as a spreadsheet, with batches, subjects and teacher names in columns. Ask for the spreadsheet directly instead of a text table.
Keep effort at Medium for routine work like this. Save High for tougher jobs, such as analysing a term’s results across many batches. For more ideas, see how coaching centres can use AI.
Digital marketing agencies
An agency can share a sample of a client’s past posts as its house style, then ask for a month of Instagram captions in Hinglish. For landing page drafts, name the design patterns the client dislikes rather than saying “make it look premium”.
Agencies building chatbots for clients on the API should read the developer guide, as some older settings, such as turning thinking off, now return an error on Opus 5.5.
How to try it
- Open claude.ai or the Claude app and sign in.
- Click the model name next to the send button and check that Claude Opus 5.5 is selected.
- Open the same menu, choose Effort and keep it on Medium to start.
- Review your saved prompts, projects and custom instructions. Delete lines such as “think step by step” or “think carefully”.
- Write your next request in one message: goal, audience, format, length and when to ask you questions.
- Attach any chart, bill or table as an image or file instead of typing it out.
- If the answer is not good enough, raise effort to High before switching to another model.
Try this prompt
I run a small home bakery in Indore that sells cakes and cookies on Instagram and WhatsApp. I have attached a screenshot of last month's orders by product and day.
Create a one-page report as a document for me, the owner. Include:
1. The three best-selling products and the slowest day of the week.
2. Three simple actions to raise sales next month, each under 30 words.
3. A short WhatsApp broadcast message in Hinglish for a weekend offer.
Use simple English. If any number in the screenshot is unclear, ask me before writing the report.
Want more like this? Our list of ChatGPT prompts for business owners in India also works well in Claude. More ready-to-copy prompts: free AI prompt library.
Limitations to keep in mind
- The speed and quality claims come from Anthropic’s own testing. Independent results may differ.
- Higher effort and bigger models use plan limits faster, so Max effort for every chat can end your quota early.
- It is not confirmed whether free users get Opus 5.5.
- Rupee prices in news reports vary slightly, so confirm on the official pricing page.
- Like any AI model, it can still make mistakes. Check numbers, legal details and tax figures before you use them.
- Avoid pasting customer phone numbers or Aadhaar details into any chatbot; read is AI safe for Indian business data before sharing files.
Alternatives
- ChatGPT (OpenAI): Popular assistant with free and paid plans and strong file and image support.
- Google Gemini: Works closely with Gmail, Docs and Sheets, useful for teams already on Google Workspace.
- Claude Sonnet 5.5: Anthropic’s cheaper model, suited to clear, well-defined tasks.
FAQ
Do I still need to write “think step by step” for Claude Opus 5.5?
No. According to Anthropic’s 12 November 2026 guidance, Claude Opus 5.5 always reasons on its own and the effort setting is the main control. Anthropic found that removing “think carefully” lines made replies start sooner with no clear drop in quality, so you can delete them from saved prompts.
Which effort level should I use in Claude?
Start with Medium, which is the default in the Claude app. Anthropic suggests moving to High only when Medium gets stuck, and using Extra high or Max only when you see a real gain, because higher effort uses plan limits faster. Try a higher effort before switching to a bigger model.
Is Claude Opus 5.5 free in India?
Anthropic says Claude Opus 5.5 is the default model on the Pro, Max, Ultra and Team plans. Whether the free plan includes Opus 5.5 is not confirmed. Free users in India can still sign up at claude.ai and try Claude, but heavy daily work usually needs a paid plan such as Claude Pro.
How much does Claude Pro cost in India?
Indian news reports from July 2026 put Claude Pro at about ₹2,000 per month including GST, with reports varying slightly between ₹1,999 and ₹2,033. Max 5x is listed at ₹11,999 and Max 20x at ₹23,999 per month. Always check the official claude.com pricing page before paying, as prices can change.
Is Claude Opus 5.5 better than ChatGPT for small business work?
There is no single winner, and Cracckk has not run its own comparison. Anthropic says Opus 5.5 is strong at reading charts, building finished files and avoiding wrong figures. ChatGPT has a larger free plan and wide app support. Many Indian small businesses try both on the same real task before paying.
Sources
- Anthropic: Using Claude Opus 5.5 as your daily driver for knowledge work
- Claude Docs: Prompting Claude Opus 5.5
- Claude Docs: Pricing
- The Decoder: Anthropic says outdated prompting habits are slowing down Claude Opus 5.5
- Business Today: Anthropic introduces India pricing for Claude AI plans
Featured image: Warm by Travis Isaacs, CC BY (via Openverse)
