# Max Mode for Claude 3.7 - Out Now!

**URL:** <https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698>\
**Category:** Release Discussions\
**Created:** [March 17, 2025, 9:53pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698 "2025-03-17T21:53:00Z")\
**Posts on this page:** 20\
**Page:** 1

<div class="post-metadata">

**Author:** ![danperks](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/danperks/32/19087_2.png) [@danperks](https://forum.cursor.com/u/danperks)\
**Post date:** [March 17, 2025, 9:53pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/1 "2025-03-17T21:53:00Z")

</div>

> **TL:DR**
> 
> - 🧠 Has Claude 3.7 Thinking at it’s core
> - 📚 Uses the whole 200k context window of the model
> - 🛠 Has a very high tool call limit
> - 🔍 Can read more code at once
> - 💰 IMPORTANT: Only available via usage pricing, costing $0.05 per prompt **AND** $0.05 per tool call!
> - ❌ Likely **not** for the average user

As many of you have realized in your own usage, Claude 3.7 is far from just a minor upgrade from Claude 3.5.

It’s fundamentally a different model with a very different way of working to its predecessor. It’s intelligent but also highly creative and can succeed where other models fail. However, it also does a lot better with larger context than any model we’ve tested in Cursor so far.

To unlock the potential of this new way of working, we’ve been quietly cooking and can now announce max mode for Claude 3.7 is now available to users on the latest versions of Cursor 🙌

💪 it is our more powerful setup for Claude 3.7. With the Claude 3.7 thinking model at its core, with it selected in your model dropdown, no cost will be spared in completing your prompts and edits!

🧠 This new option is capable of more than any of Cursor’s existing models, combining the intelligence of Claude 3.7 with the contextual gathering of Cursor, to unlock edits and capabilities beyond what has been possible so far.

 ![Screenshot 2025-03-17 at 21.58.42](https://us1.discourse-cdn.com/cursor1/original/3X/e/d/edf1821dd5b9d8250b03741f2635a66287a6f654.png)

Behind the scenes, we push Claude 3.7 to it’s maximum!

📚 Firstly, the **entire 200k context window** is available and used, so you can fit as much of your code into your prompts as possible. Previous models have performed poorly with such tasks, but Claude 3.7 is the first model to perform better with such large context windows.

🔗 Secondly, we’ve **bumped the tool call limit** up to 200 in one chain, meaning it can make massive edits and gather context across your entire codebase in one go!

🔍 Finally, it can read much larger chunks of your code at once, meaning it needs fewer tool calls to learn what it needs to.

 ![Screenshot 2025-03-17 at 22.08.28](https://us1.discourse-cdn.com/cursor1/original/3X/1/0/102a3319a875d16e9d36c03601ec6279dee27327.png)

It is a ‘no expense spared’ feature and, as such, comes with intelligence and capabilities no other model has. **However** , the mode is expensive (see below) and is likely not worth the cost for the majority of users.

This model is our best choice for implementing large and complex projects all at once or for completing intricate code edits that require a deep understanding of their functionality to maintain them.

But for standard code edits, Cursor’s agent is still the most cost-effective solution and would be more than enough for \>90% of all the prompts Cursor sees.

We only recommend this model to those who can afford to test its capabilities without concern about the cost and will make use of its additional intelligence.

 ![Screenshot 2025-03-17 at 21.59.00](https://us1.discourse-cdn.com/cursor1/original/3X/6/0/60a6087f02e4163c34fe41df3c4f9b3e5b2e380d.png)

All these changes come at a cost, and as such, usage of Max mode for Claude is not available within the base Pro plan.

Instead, the mode is **only available by enabling usage-based pricing**.

**‼ Max mode Claude 3.7 prompts cost $0.05 per request + $0.05 per tool call ‼**

⚠ This does mean that, if left unchecked, it would could cost $10 if left to loop through all 200 tool calls it has available.

However, it is the most powerful model offering we’ve had in Cursor to date, and we believe for many advanced users, the cost will be more than worth it in the productivity increase it gives in cases where models have failed in the past.

---

<div class="post-metadata">

**Author:** ![jake](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/jake/32/20459_2.png) [@jake](https://forum.cursor.com/u/jake)\
**Post date:** [March 18, 2025, 9:48am UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/9 "2025-03-18T09:48:30Z")

</div>

This is awesome. Super pumped to take this for a spin.

I think it would help if a “tool call” was clearly defined.

As while tool calls are sometimes obvious—such as MCP tool calls—what about things such as terminal commands or code grepping?

---

<div class="post-metadata">

**Author:** ![JonoGee](https://avatars.discourse-cdn.com/v4/letter/j/3da27b/32.png) [@JonoGee](https://forum.cursor.com/u/JonoGee)\
**Post date:** [March 18, 2025, 10:12am UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/10 "2025-03-18T10:12:43Z")

</div>

Excellent!!

I must say that for the last 2-3 Days Claude3.7-Thinking has been working very well!!

Whatever you have been doing its working!! Well Done! Now just make sure updates don’t ruin all the hard work before releasing them!

---

<div class="post-metadata">

**Author:** ![JonoGee](https://avatars.discourse-cdn.com/v4/letter/j/3da27b/32.png) [@JonoGee](https://forum.cursor.com/u/JonoGee)\
**Post date:** [March 18, 2025, 10:23am UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/11 "2025-03-18T10:23:58Z")

</div>

How is this possible if a few days ago everything was at complete and utter overloaded capacity though?

---

<div class="post-metadata">

**Author:** ![kstars111](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/kstars111/32/14862_2.png) [@kstars111](https://forum.cursor.com/u/kstars111)\
**Post date:** [March 18, 2025, 10:37am UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/12 "2025-03-18T10:37:48Z")

</div>

What is a tool call?

---

<div class="post-metadata">

**Author:** ![john-savepoint](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/john-savepoint/32/15998_2.png) [@john-savepoint](https://forum.cursor.com/u/john-savepoint)\
**Post date:** [March 18, 2025, 11:21am UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/13 "2025-03-18T11:21:08Z")

</div>

Have you fixed the double charging of your customers yet? I’m still waiting to hear back from anyone after several weeks of trying to get in touch. 4 months of double charging. I could afford alot of Claude 3.7 Max if I had the $80 you overcharged back.

---

<div class="post-metadata">

**Author:** ![ailee](https://avatars.discourse-cdn.com/v4/letter/a/e5b9ba/32.png) [@ailee](https://forum.cursor.com/u/ailee)\
**Post date:** [March 18, 2025, 11:38am UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/14 "2025-03-18T11:38:28Z")

</div>

After several paying months you “forced” me to make an account here.

Let me clarify what you did here:

1. Sonnet 3.7 is not to good in Cursor (you know where and why).
2. You don’t fix it.
3. You introduce fixed version outside subscription.

Correct?

---

<div class="post-metadata">

**Author:** ![jsboer](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/jsboer/32/35838_2.png) [@jsboer](https://forum.cursor.com/u/jsboer)\
**Post date:** [March 18, 2025, 11:50am UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/15 "2025-03-18T11:50:40Z")

</div>

They must consider the cost. They need to earn money. So the MAX comes out, which can assure most people afford the subsribe while someone can use MAX for more hard job by spending more money.

---

<div class="post-metadata">

**Author:** ![jake](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/jake/32/20459_2.png) [@jake](https://forum.cursor.com/u/jake)\
**Post date:** [March 18, 2025, 11:56am UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/16 "2025-03-18T11:56:54Z")

</div>

3.7 works fantasticly well in Cursor for me. I’ve stopped using 3.5 entirely.

The Cursor team is continuing to work on the integration of 3.7.

While 3.7 Max takes advantage of 3.7, it uses it in a very different way that uses a lot more compute and requests. Hence the usage-based pricing.

---

<div class="post-metadata">

**Author:** ![danperks](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/danperks/32/19087_2.png) [@danperks](https://forum.cursor.com/u/danperks)\
**Post date:** [March 18, 2025, 11:56am UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/17 "2025-03-18T11:56:57Z")

</div>

@jake @kstars111 Thanks for the points about tool calls. I’ll add this to the docs today, but to summarise, a tool call is any action the AI decides to take outside of writing it’s own output. This does include reading and editing files, searching the web, and grepping code. Terminal commands only count if the Agent runs them inline, as it monitors the output itself - if it just suggest a terminal command, this is not a tool call!

@JonoGee Great to hear things have been working well for you recently! We’ve been working hard to resolve the capacity issues, mainly by distributing the load across multiple Claude hosters, such as Anthropic themselves, and AWS Bedrock. As you’ve mentioned, we should be at a much better place now than we were a week ago!

@john-savepoint I don’t believe we have any ongoing issues with being charged twice as you describe. I know we’ve spoken already on this, and I’ll chase up your email now and get this sorted for you!

@ailee As described in the initial post, Claude 3.7 is built for a different way of working to Claude 3.5 - it is not simply just an upgrade on the last model. Claude 3.7 works best in more expensive environments (like Claude Code!), and while we have worked hard to make its performance comparative to our other models, we wanted to ensure we didn’t lock away any of Claude 3.7’s ability just because it works differently to our other models.

As Jono says above, a lot of users are now happy with Claude 3.7’s performance, and we are, of course, still tweaking and refining it to try and iron out any weird behaviors. However, Max mode for Claude runs at a cost that could not be included in the Pro plan, so there was never an option to include such a setup in the base plan, irrespective of whether we called it “Max mode for Claude”, or just `claude-3.7-thinking`.

---

<div class="post-metadata">

**Author:** ![QTom](https://avatars.discourse-cdn.com/v4/letter/q/b77776/32.png) [@QTom](https://forum.cursor.com/u/QTom)\
**Post date:** [March 18, 2025, 12:21pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/19 "2025-03-18T12:21:18Z")

</div>

Have you resolved the issue of the GitHub Copilot chat window disappearing in version 0.47?

---

<div class="post-metadata">

**Author:** ![QTom](https://avatars.discourse-cdn.com/v4/letter/q/b77776/32.png) [@QTom](https://forum.cursor.com/u/QTom)\
**Post date:** [March 18, 2025, 12:23pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/20 "2025-03-18T12:23:34Z")

</div>

Disabling GitHub Copilot Chat is for 2 times 3.7 thinking and the Max model?

---

<div class="post-metadata">

**Author:** ![qwert](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/qwert/32/48940_2.png) [@qwert](https://forum.cursor.com/u/qwert)\
**Post date:** [March 18, 2025, 12:38pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/21 "2025-03-18T12:38:18Z")

</div>

Fix linter error also counts as tool call, this seems unreasonable, what do you think

---

<div class="post-metadata">

**Author:** ![ai-nuke](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/ai-nuke/32/30267_2.png) [@ai-nuke](https://forum.cursor.com/u/ai-nuke)\
**Post date:** [March 18, 2025, 12:43pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/22 "2025-03-18T12:43:17Z")

</div>

I was dealing with a very annoying UI bug which I tried to fix with Cursor, Wind$urf, Claude code AND Cline where all failed repeatedly. 3.7 Max mode fixed it in one go!

---

<div class="post-metadata">

**Author:** ![Erik1](https://avatars.discourse-cdn.com/v4/letter/e/8e8cbc/32.png) [@Erik1](https://forum.cursor.com/u/Erik1)\
**Post date:** [March 18, 2025, 12:45pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/24 "2025-03-18T12:45:14Z")

</div>

Terrible, holy. Chewed thru 150 credits last night- DID NOTHING but give errors, single page html and css with a funky api call. I am kind of glad I am out of credits for the month- don’t like being dependent on a single service, cursor is so mercurial, I will try surfing 😉

---

<div class="post-metadata">

**Author:** ![sdfgsdfgd](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/sdfgsdfgd/32/34748_2.png) [@sdfgsdfgd](https://forum.cursor.com/u/sdfgsdfgd)\
**Post date:** [March 18, 2025, 1:09pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/25 "2025-03-18T13:09:21Z")

</div>

why did you remove the 3.7-thinking model , leaving people with only a thinking model that costs per use ? is this an intentional change ??

I only have 3.7-sonnet and 3.7-sonnet-max, 3.7-sonnet-thinking is gone ☹ what did these people do to deserve this

---

<div class="post-metadata">

**Author:** ![Maxence](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/maxence/32/37854_2.png) [@Maxence](https://forum.cursor.com/u/Maxence)\
**Post date:** [March 18, 2025, 1:10pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/26 "2025-03-18T13:10:58Z")

</div>

@sdfgsdfgd You can uncheck “thinking” mode to use 3.7 without thinking

---

<div class="post-metadata">

**Author:** ![sdfgsdfgd](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/sdfgsdfgd/32/34748_2.png) [@sdfgsdfgd](https://forum.cursor.com/u/sdfgsdfgd)\
**Post date:** [March 18, 2025, 1:11pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/27 "2025-03-18T13:11:40Z")

</div>

I meant, in the settings there isn’t even the thinking model present anymore, after the update. I can’t enable or disable it. Cursor Settings → Model Names doesn’t have 3.7-thinking model anymore.

edit: wait I finally found the switch, you are right xD

---

<div class="post-metadata">

**Author:** ![leoing](https://avatars.discourse-cdn.com/v4/letter/l/dec6dc/32.png) [@leoing](https://forum.cursor.com/u/leoing)\
**Post date:** [March 18, 2025, 1:26pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/28 "2025-03-18T13:26:40Z")

</div>

![image](https://us1.discourse-cdn.com/cursor1/original/3X/4/f/4f044fa726b316b88b37bace8115b83dcc61252f.png)

---

<div class="post-metadata">

**Author:** ![ParkerRex](https://sea3.discourse-cdn.com/cursor1/user_avatar/forum.cursor.com/parkerrex/32/7169_2.png) [@ParkerRex](https://forum.cursor.com/u/ParkerRex)\
**Post date:** [March 18, 2025, 2:31pm UTC](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698/29 "2025-03-18T14:31:03Z")

</div>

**1. Codebase Interaction Tools:**

• **read\_file(full\_path: str)**: Reads the content of a specified file.

• **write\_file(full\_path: str, content: str)**: Writes or updates content in a specified file.

• **delete\_file(target\_file: str)**: Deletes a specified file from the codebase.

**2. Code Modification Tools:**

• **edit\_file(target\_file: str, instructions: str)**: Edits a file based on provided instructions.

• **parallel\_apply(edit\_plan: str, edit\_files: list)**: Applies similar edits across multiple files simultaneously.

**3. Code Search and Analysis Tools:**

• **codebase\_search(query: str, target\_directories: list)**: Finds code snippets relevant to a search query within the codebase.

• **grep\_search(pattern: str, target\_directories: list)**: Performs a grep search using a specified pattern.

• **file\_search(filename: str)**: Searches for a specific file by name.

• **web\_search(query: str)**: Conducts a web search for external information.

**4. Code Execution and Debugging Tools:**

• **run\_command(command: str)**: Executes a command in the terminal.

• **reapply(target\_file: str)**: Reapplies the last edit to a specified file using a more advanced model if necessary.

These tools work together to provide a seamless and efficient coding environment within Cursor IDE.

[Next page](https://forum.cursor.com/t/max-mode-for-claude-3-7-out-now/65698.md?page=2)
