Hacker News new | past | comments | ask | show | jobs | submit login

I don't have ads on my SearXNG instance and I find that its comparable to Kagi. Have you tried searching with SearXNG?



I mean, it's something I'd consider. I have a little ARM SBC here that I run Plex and Immich etc on. But Kagi makes it very easy for a not terrible price -- not much more than the monthly power bill for a little server -- and they also do more than just grab from other search services as they _do_ have their own crawler (unclear how much it's used for results?) and have LLM stuff which isn't terrible.

Also as a former Google employee... I dig the idea of contributing funds for some competition in this space.


How much maintenance does running a private instance take, and would you say it should be fine to run from your own IP? I've been using Kagi for ages, but I wouldn't mind a fallback that does better than DDG.


Not much? I mean, I set it up initially to use as an LLM search tool for agents. That required me to use a specific version of SearXNG due to a JSON bug. My exact docker command is here:

    docker run -d -p 9017:8080 --name searxng_data -v "${PWD}/searxng_data:/etc/searxng" -e "BASE_URL=http://localhost:9017/" searxng/searxng:2024.8.31-769b2ed03

I run this from the directory where I want to store the data on windows (hence the weird PWD thing). I also use cloudflare tunnels and have the internet facing website, https://sear.mydomain.com, pointed at http//localhost:9017. I set up all my cloudflare tunnels with email authentication. My token lasts a month (but is configurable to whatever time frame you like). My traffic is strictly mine, so it seems to be working ok.

Its basically no maintenance after this setup. I can mess with all the settings (except results format) from the main interface.


I appreciate the writeup! I'll likely go down this rabbit hole soon.


Why go though all that hassle when you can easily use Kagi?


Because I can configure exactly which search engines to index, no ads, use it for my local LLM's for search, better image search (maybe not better, but totally different results), a bazzillion tweaks and mods. I don't have to pay Kagi for the same results or keep track of how many searches I have left.


What is the hassle? I set it up once and just use it. When I hear about search engines that I don't think I want in my mix I go in and turn it off.


I looked into it before Kagi existed and you needed to be a sys admin or developer to be able to use it. It's not for the general public.


Not at all. You can run it with docker or you can use one of the myriad public instances.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: