1× cost = $10, 1× Sonnet 4 usage = 300 msgs The base accounting unit is Copilot Pro ($10 for 300 Sonnet 4 msgs)

Damn, I’m not even hating right now. It’s SOOO bad.

I’ve asked Claude to make a visualization for me and I was thinking of adding this to the original apology at last? post but nahh, I think this visualization deserves its own post.

The key insights are:

• For however much you pay, you should get at least 5% more usage than what you pay for the plan to be competitive considering the entire AI-assisted tool landscape.

  • At 5× the cost, expect 6% more usage (5.3× usage over 5× cost)
  • At 10× the cost, expect 10% more usage (11× usage over 10× cost)
  • At 20× the cost, expect 13.5% more usage (22.7× usage over 20× cost)
  • Know that clearly when you don’t get more usage at $50 USD per mo. and above, you’re incurring a loss. It might be more acceptable for Augment Code because, their pricing is more straightforward and they’re a smaller company. See Note [5] below
  • Whether Cursor or any other provider incurs losses is irrelevant to users, and their survival as a company is not my business

• Only Augment Code plans are about as “uncompetitive” as Cursor’s plans.

  However, you can spend $10 for 100 more messages, while for Cursor you go token-based, and you get charged for tool calls and auto context attachments. So that’s where any additional spend beyond Cursor base subscription can quickly outrun Augment’s.

  • Also, Augment Code plans count messages, not tokens. See Note [5] below

• Trae Pro, Anthropic Claude Pro/Max plans, and Microsoft Copilot Pro/Pro+ plans offer great value at $10, $20/$100/$200, and $10/$39 price points, so we do have a lot of great alternatives.

• This visualization does not capture additional benefits beyond Claude Sonnet 4 usage.

  • Copilot Pro/Pro+ offers unlimited GPT-4.1 access even in agentic mode;
  • Trae Pro has “unlimited number of slow requests” (sound familiar?)
  • Anthropic’s Claude subscription offers Claude chat (claude.ai), artifact creation, app publishing, and a multi-agent research feature.
  • Cursor has unlimited Tab completion I guess, but this is not the moat you think it is

• You’re coping for thinking Cursor is competitive pricing-wise. You might be able to make a case for Tab Completion, but it’s not worth $20/mo. ∎

Note

1. The basic accounting unit is GitHub Copilot Pro, which costs $10 USD per month, giving 300 premium messages, which allow the use of Anthropic Claude Sonnet 4

   (as well as other models, including OpenAI o3 and Google Gemini 2.5 Pro, but they’re not what catalyzed the Cursor pricing model change, as we had established)

2. The graph uses logarithmic scale as opposed to linear scale to capture the full 0.75-133.33 range within the same graph meaningfully

3. Cursor usage figures are based on "median token usage”

   • Source: https://docs.cursor.com/account/pricing
   • The content of that docs page may change without notice by the time you visit it — check the time of creation for this post and use Wayback Machine to find the data that led to the creation of the visualization

4. Anthropic Claude plan figures are based on minimum and maximum number of Claude Code prompts a user is able to sent per 5-hour session

   • Claude Pro plans may or may not have the same 50-session restriction. We don’t know, or at least I couldn’t find it. So I had to assume so.
   • Most users should be allowed to use at least the minimum number of Claude Code prompts, while only the least demanding, simplest prompts can reach towards the maximum. However, even if we only consider the minimum figures, Claude Pro/Max for Claude Code with Sonnet 4 usage is still highly competitive.
   • Source: https://support.anthropic.com/en/articles/11145838-using-claude-code-with-your-pro-or-max-plan

5. It might seem like Cursor’s new Pro+ and Ultra plans try to one-up Augment Code’s Pro and Max plans, but Augment Code actually switched to a simpler message-based pricing model on May 7, 2025, where tool calls etc. are not additionally billed: https://www.augmentcode.com/blog/new-simpler-pricing-with-user-messages

   • Windsurf similarly removed “Flow Action credits” even earlier on Apr 21, 2025: https://windsurf.com/blog/pricing-v2