The Pulse: Interesting AI coding stats from Cursor

A recent report from Cursor reveals that power users generate 10x as many lines of code as the median, with the top 1% of users creating around 30-40K lines of code per week. The report also shows that Cursor consumes 10x more input tokens than it generates in output tokens, with 90% of token usage being input tokens. This highlights the importance of caching context to reduce token costs, with Cursor's caching mechanism reducing token costs by 10x. The practical implication for engineers building AI systems is to prioritize context reuse and caching to improve efficiency.

⚡ Key Takeaways

• The top 1% of Cursor users generate around 30-40K lines of code per week, equivalent to what ~45 "median" devs generate in the same period.
• Cursor consumes 10x more input tokens than it generates in output tokens, with 90% of token usage being input tokens.
• Input tokens are priced at a fraction of output tokens, but still account for closer to 70% of the cost of AI coding agents due to their high usage.
• Cursor's caching mechanism reduces token costs by 10x by avoiding re-generating old context with new input tokens.

The findings from Cursor's report highlight the importance of optimizing AI token usage and caching context to reduce costs and improve efficiency. This has significant implications for engineers building AI systems, particularly in workflows like coding where a lot of existing context is reused.

✅ Practical Steps

1. Implement caching mechanisms to reduce token costs and improve efficiency in AI coding workflows.
2. Optimize AI token usage by minimizing output tokens and maximizing input token reuse.
3. Consider using AI coding agents like Cursor that prioritize context reuse and caching to reduce costs.

Want the full story? Read the original article.