←back to thread

Ask HN: Cursor or Windsurf?

310 points skarat | 4 comments | 12 May 25 04:41 UTC | HN request time: 0.668s | source

Things are changing so fast with these vscode forks I m barely able to keep up. Which one are you guys using currently? How does the autocomplete etc, compare between the two?
Show context
danpalmer ◴[12 May 25 05:28 UTC] No.43959899[source]▶
Zed. They've upped their game in the AI integration and so far it's the best one I've seen (external from work). Cursor and VSCode+Copilot always felt slow and janky, Zed is much less janky feels like pretty mature software, and I can just plug in my Gemini API key and use that for free/cheap instead of paying for the editor's own integration.
replies(9): >>43960069 #>>43960506 #>>43960546 #>>43961423 #>>43961614 #>>43962057 #>>43962974 #>>43967753 #>>44019166 #
wellthisisgreat ◴[12 May 25 06:01 UTC] No.43960069[source]▶
Does it have Cursor’s “tab” feature?
replies(2): >>43960155 #>>43961890 #
1. dvtfl ◴[12 May 25 06:21 UTC] No.43960155[source]▶
Yep: https://zed.dev/blog/edit-prediction
replies(1): >>43960299 #
2. eadz ◴[12 May 25 06:53 UTC] No.43960299[source]▶
It would be great if there was an easy way to run their open model (https://huggingface.co/zed-industries/zeta) locally ( for latency reasons ).

I don't think Zeta is quite up to windsurf's completion quality/speed.

I get that this would go against their business model, but maybe people would pay for this - it could in theory be the fastest completion since it would run locally.

replies(2): >>43961605 #>>43962492 #
3. xmorse ◴[12 May 25 10:48 UTC] No.43961605[source]▶
Running models locally is very expensive in terms of memory and scheduling requirements, maybe instead they should host their model on the Cloudflare AI network which is distributed all around the world and can have lower latency
4. rfoo ◴[12 May 25 13:06 UTC] No.43962492[source]▶
> the fastest completion since it would run locally

We are living in a strange age that local is slower than the cloud. Due to the sheer amount of compute we need to do. Compute takes hundreds of milliseconds (if not seconds) on local hardware, making 100ms of network latency irrelevant.

Even for a 7B model your expensive Mac or 4090 can't beat, for example, a box with 8x A100s running FOSS serving stack (sglang) with TP=8, in latency.