“https://www.harness-bench.ai/leaderboard.htmlThis provides a method, but the data looks stale and perhaps a bit thin compared to say, Cursor, or even AntiGravity data.”
“The first party cursor models (initially composer and now grok-cursor) and price very competitively for their quality and have very generous usage limits included in the base plans, and then per token pricing afterwards.”
“The nice part about Cursor to me is you can write your own specs, have the ai maintain it and write their own plans, and see it all right in front of you.”
“I've started running two Cursor agents on the same repo, one on a feature, one cleaning up an old module. It works fine until they touch the same file.”
“First it was ChatGPT. Then Claude. Then Cursor. Then Claude Code. Then Gemini. Then Codex. Then a bunch of indies. Then Claude still. Now Grok and Cursor's back.”
“I’ve used Grok 4.6 on Cursor, and it feels almost unlimited at the moment. That too on a $20 sub! It’s a good model for regular day-to-day use, which is how the majority of people use AI at work.”
“Cursor just moved the bottleneck to App Store review. I automated metadata uploads and build number syncing so CI handles everything outside the repo”
“Cursor stays quiet because real users ship code instead of posting reviews. Mainstream channels cover products that pay for placement or generate engagement metrics”
“I'd say Cursor is my top pick compared to the others. I use Cursor with the Composer model, and it really does what I need with very minimal post-adjustment. This is especially true when I compare it to other AI agents and tools.”
“Cursor can now host your code. Cursor Origin is a Git forge with repositories, pull requests, agents, automations, GitHub mirroring, a CLI, and an API.”
“In my opinion, cursor is better for programming because when it responds, it shows a list of changed files and lines of code, allowing you to review, revert, or accept each line or file individually.”
“Can't edit: Looks like they edited this page after the fact[1]: It directly said:> Continuity (cnt for short) is the git storage system [...]I guess someone realized that turned this into even more of a joke than it already is.[1]: https://web.archive.org/web/20260818210312/https://cursor.co...”
“I had so many bugs in the last couple of weeks on the daily level, that it feels like Cursor is not being tested very well (I hope they don't vibe-code parts of it). It's frustrating to use.”
“Cursor is great until you hit that fast request cap halfway through the month and your momentum completely dies. I ended up switching my main workflow to Freebuff to take advantage of their 3+ daily free premium hours instead.”
“Cursor feels polished but the agent mode still does weird stuff sometimes, like refactoring files I never asked it to touch. Haven't used codex enough to have a strong take on it.”
“But yet, I always come back to Cursor, I just really like it, feels good for me and has great co text management out of the box. Shame it doesn’t always work well with local models.”
“i still work with a small number of agents at a time, but primarily switched to run things with cloud agents + cursor ios when im not working from home. also don’t have to deal with worktrees locally is a huge plus.”