Mullvad is deprecating their google search proxy, Leta

  • 🇵🇦 Nuestro primer dominio localizado está en español en kiwifarms.pa. Our first localized domain is on Spanish on kiwifarms.pa.
  • Want to keep track of this thread?
    Accounts can bookmark posts, watch threads for updates, and jump back to where you stopped reading.
    Create account

skunt

Morning! Have a ship!
True & Honest Fan
kiwifarms.net
Registrado
3 de Dic, 2023

On November 27, 2025, we will shut down our search proxy, Leta.


Leta's primary benefit was that it acted as a privacy proxy for search. Pooling and caching requests on behalf of a great number of users.


The search industry continues to undergo big changes. Leta will not be able to follow and will likely become less useful over time.


Similar privacy can be achieved through the combination of a VPN and a privacy-focused browser.


We have therefore decided to discontinue Leta and continue to advance the development of state-of-the-art of VPNs and browser privacy - through our own work and research and in collaboration with our partners.

Bummer, I made extensive use of it. Anyone else using it?
Thinking i'm forced to setup my own instance of searxng or something https://github.com/searxng/searxng. Kagi is supposed to be decent but I'm not paying for search. I used to use DDG and it wasn't bad.
Anyone have any suggestions?
 
Anyone else using it?
Nope, seems like one of those things that a handful of people used, not enough to justify it's upkeep, and once it inevitably got closed the handful of people that used it got upset. Same thing happened with Mozilla Pocket, if you can remember such a thing existing.
Thinking i'm forced to setup my own instance of searxng or something
The issue with self-hosting SearX is that it defeats the idea of having a search proxy. The idea is that you have multiple users querying the search engines from a single instance where all the upstream search engines see is the same SearX server, leading to the harvested data becoming useless, since it's a lump of multiple individual searches with no way to differentiate between them. Once you're running a local instance that only you are using, especially if it's on your home network with the same front facing IP, then you once again become easy to identify by Google, Bing and so on.
 
The issue with self-hosting SearX is that it defeats the idea of having a search proxy.
Not a problem really, I already host a bunch of services for ~20ish people. I just never really wanted to touch search because I didn't need to. I think I'd still prefer not to if there's better options out there, I guess it's a last resort.
edit; actually i just realised that probably 0 of them would go to the effort of resetting their default search so yeah, probably not worth it.
Nope, seems like one of those things that a handful of people used
It's been having issues for a few weeks which probably prompted it.
 
I use brave as default and an LLM (normally grok) if what i'm looking for is more obscure or hard to phrase for a "dumb" search.
 
I don't use search engines anymore, I just guess URLs. I've gotten pretty good at it too
 
I try, but Brave's search results suck so bad and its AI dealie might actually be MORE annoying and cucked than Google's.
I think both the search and the AI have gotten better over time, although it can be hard to tell and maybe I'm just used to its quirks now. I must have been using it since mid-2024 based on my KF posts.

I have done a small amount of vibe coding with the Brave Search AI, basically just describing a function or regular expression I want. I think it's also returning more factual answers in response to search queries, by citing sources from the search results. My feelz are confirmed by this recent blog post:

September 29, 2025: Introducing Ask Brave (archive)
AI answers are not new in Brave Search—we released the first version of this feature in March 2023 (known as “Summarizer”). Today’s version—AI Answers (originally Answer with AI)—serves over 15 million answers per day, making it one of the most widely used AI products in the market. Ask Brave is the next iteration of this experience, combining the best of both worlds: the simplicity and direct action of the conventional Brave Search engine, and the convenience and time-saving benefits of LLM-generated responses. Ask Brave is available in addition to AI Answers, which are meant to provide users with quick answers, while Ask Brave gives users more extensive results. Users can easily follow up with more queries in chat fashion, thus combining search and chat.

The answers provided by Ask Brave are always grounded in information found on the Web, meaning that the AI is less likely to hallucinate or provide irrelevant information. As a search engine, we ensure the AI limits itself to search results that are relevant to your question.
Ask Brave is built on Brave’s recent grounding technology for the Search API, which achieves state-of-the-art scoring on factual accuracy benchmarks (94.9% accuracy on SimpleQA). This Deep Research technique leverages multiple rounds of Brave’s proven search against our junk-free index of 35+ billion webpages. The system has been designed to iteratively cover blindspots, issuing dozens of queries and analyzing thousands of pages so users receive the most thorough and well-informed responses available.
 
I think both the search and the AI have gotten better over time, although it can be hard to tell and maybe I'm just used to its quirks now. I must have been using it since mid-2024 based on my KF posts.

Even in Brave Search's infancy, it was infinitely superior to DuckDuckGo. I often found that Brave Search struggles with semantics, i.e. looking up exact quotes for memes, their search operator syntax is opaque, if not outright nonexistent (only using quotation marks, no +term no -term, etc), but honestly? It was still pretty damn good even then. I also found that it was helpful when scouring for dirt to look up on lolcows that Google and Bing overlooked. I also appreciate how you don't need to rely on !bangs to redirect to Google or Bing. If you scroll a little bit down and you don't find what you're looking for, you can easily click the "Google," "Bing," or even the fucking "Mojeek" button to get a clean redirect.

Brave Search's local results are also more than adequate, having improved considerably with time. If I narrow down my location cursor to my house, and I search "barbers near me," I don't get barbers where my ISP's data centre is. I get barbers that are literally within walking distance of me. It's still not as pretty as Google or Bing where they have the nice little cards that consolidate all the contact info including website and phone number, but I don't necessarily need that if I'm just looking for ideas. I think Brave Search has a huge leg up over DuckDuckGo because it's a wholly independent search engine with a unique crawler, instead of being a rehash of Bing like DuckDuckGo was (RIP Yahoo being independent, RIP Yandex integration). It ain't perfect, but it gets the job done without me groaning in frustration anywhere near the amount of times I did with DDG.

The crypto integration with Brave the browser forever tainted its reputation in my eyes, but the search engine's still pretty damn ace.
 
Kagi is supposed to be decent but I'm not paying for search.
It is good and you should. Seriously though, Internet searches are the single most frequent thing most people do online. If you're searching 5-10 times per day, every day of every week, it's worth paying a cup of coffee a month for something that isn't biased and isn't data-mining you. (Inb4 "I grow my own coffee beans and heat the water in my solar farm").

Businesses paid about $200bn to search engines annually. That's how powerful their search manipulation and advertising is perceived to be. Search results are one of the basic elements of the Web. Eh, I don't have to convince anybody to hate Google. But Kagi is worth a shot, imo. I talk about it here from time to time - it has some very nice features, too.
 
Atrás
Top Abajo