← Back to context

Comment by threePointFive

6 days ago

Can someone comment on the cost of running agentic models? Not for a company but for an individual. I tried "vibe coding" a personal project I was struggling with and left even more frustrated because I kept running into token rate limits with Claude (used inside of Zed if it matters). Did I pick the wrong model, the wrong editor, or do I just need to not be so tight with my money?

I tried Zed's agent support with Copilot. I was able to get it to implement a working feature with tests, but the code quality was poor and it took longer to do it than if I had written it myself.

I am wondering if maybe the average programmer is way slower and worse than I thought.

I haven't used zed specifically, but were you using your own API key for Claude? If so, you were probably running into the Anthropic API rate limits [1]. You can either a) deposit more funds to move your way up the tier list, or instead access Claude via something like OpenRouter, which will give you much higher limits.

[1] https://docs.anthropic.com/en/api/rate-limits

Try Cursor.ai or Windsail. They have free trials for the good models, and are pretty cheap to use if you decide you like them.