Thanks for pointing out safereddit and other redlib instances! I had the old.reddit.com redirect configured for Uruky and was devastated that it no longer worked. Updated, cheers!
Me too. The longer I use it, the more I like it. Initially I found myself going back to google fairly often, but unless I'm specifically trying to find a product for sale in my region, Kagi is almost always better now. Maybe Kagi got better, or google got worse, or probably both.
> Reddit has recently blocked access to old reddit without an account
I don't understand why this is a barrier considering their accounts aren't invasive at all since you can use a temporary email account to sign up, and then you never have to do anything via email after that.
They are pretty ban-happy, and generally outright hostile to newly created accounts that try to participate.
I created an account recently after not having one for a couple years. I was banned within 2 weeks. I accidentally connected to my VPN when I had a Reddit page open in a minimized window that I forgot about. I assume that’s what did it. Last I checked, I couldn’t even login with this account anymore.
I don't like assisting reddit's monetization/AI training machine by tying my browsing activity directly to an account identity. I would prefer to browse anonymously and make them work at least a little bit harder to compile my anonymous activity into a coherent single-user profile.
If that doesn't bother you, then good for you, and we have different feelings about privacy.
I just don’t want an account. I disagree with how they’re running things these days and I just don’t see the point when I only want to read posts, not make them, or comment or vote on them. And the more they push “Make an account! Get the app!” the less likely I am to do it. It’s a really hostile way to treat users. I don’t like most social media and it doesn’t bother me not having accounts on them, but it’s become such a pain with Reddit because it was one of the only places online left to get decent information about specific things. I’m now finding myself searching for things online and finding no answers because Reddit is inaccessible and that’s where the answer tends to be.
I'm annoyed mostly because I always surf with private browsing, so I always have to log in again and again. And for some reason I haven't bothered checking, Firefox doesn't save the username field.
NEW Reddit has recently started blocking logged-out access too. Stupid modal popup with "Join the most real place on the internet" (yeah, right) banner that appears after about 20 seconds.
I'm not logging in to my private account from a work machine that's MITMd for "security", just to browse search results or take a short break!
(I've been a user for nearly 21 years, but using it less and less with each stupid choice - these days I post a comment maybe once a week there. I think I may finally be done with Reddit, sadly. It was good while it lasted.)
> NEW Reddit has recently started blocking logged-out access too.
That'd be an important milestone in enshittification, when it comes to other entities that were happy to outsource their fan-forums, support, announcement systems, etc.
Kind of like how many groups started using Twitter as a way to provide a stream of announcements--sometimes time-critical ones--to a broad audience. The entire use-case goes down the toilet when users aren't being shown your stuff in chronological order anymore.
I know new reddit has its potential inspiration in stopping LLM scraping(?) but websites are made worsebecause of some stupid metric that has to be fulfilled
llm scraping? they encourage it, recently chatgpt stopped referencing reddit as of late and thier traffic dropped, that is why they are trying to make people login.
Hmmm, so if you are looking for some news item, you'll only find AI-written clickbait articles covered in pop-up ads, rather than any news sites that employ journalists?
That's why it is a choice for companies to not pay for (human) software engineers anymore for writing software and to get all the agents to do the work and write the software that we use today.
Just like how we will see articles now being written by AI that will rank in Kagi instead of human written articles.
Not all news sites are paywalled (reuters for e.g) and if I use a search engine I'm not looking to experience unwanted advertisement (paywalls are an ad.)
The small handful of news sites I willingly subscribe to are in my RSS reader (where stories are even more de-junked than is possible in a browser), so paywalled sites showing up in search adds no value.
If a news site wants to join the ranks of my RSS feeds they need to be of high enough quality that their reputation does more work than paywalled search results do.
I'd actually want to know if it is a paid website BEFORE clicking on it.
Nothing worse than clicking on it, and after 5 seconds you get the naggy pop-up that disables the scrolling. Now I need to send it to archive.ph to read it.
I'm fine with a paywall, but put it above, don't interrupt me as I am reading.
There's more than one model for news and some of those models don't even result in advertising-controlled biases. Public media sites employ journalists for instance.
I love Kagi but will probably be selective or ignore this one. Sometimes, I like to know what's out there, and I don't have a problem with paying for good content.
Because it lets me hide the trash that is commercialized internet. I choose what service I pay for, e.g. Kagi, everything else should just vanish from my point of view and stop polluting my eye space.
What's trash about it? Giving no explanation does not contribute to a conversation.
I also don't care if people have original ideas if they take an existing idea and do it better, which is what I appreciate about Kagi at the moment. I don't want anything interesting or novel. I want search that doesn't berate me with ads or AI overviews if I don't ask for them.
Would the biggest impact of this be for things like news sources? I guess the idea is that ads are preferrable, as ads and paywalls (subscriptions basically, right?) seem to be two of the primary existing ways that such places support themselves (or don't)?
Are we talking a company rewording reporting that a different news company hired 1 or more journalists, etc. to originally report? And in this example, are both the company paying the journalist(s) etc, and the site rewording the 1st company's content, "free" because of ads if not paywalls?
If they had a "here is what I'm currently subscribed to" that might be valuable. Everybody should subscribe to the local news paper - to get local news. Things that matter to them that they won't hear about. (Odds are you the reader do not live in the same city as me and so have no interest in who is on my local school board - but I'm trying to figure out who to vote for and need more information than I can find).
I'd be onboard with supporting a local newspaper -- if they hadn't gone all-in with aligning themselves to a particular political tribe. They're part of the problem Ground News makes obvious to a hilarious degree, where if something happens that doesn't fit the narrative you just will never see it if you only look at their platform. Given that not everyone has a reputable local newspaper I don't know that saying "everybody should" is necessarily sound advice.
Anyway, you can block sources individually in Kagi, that can weed out the most common paywall options you don't have, like Forbes or whatever.
should is the key here. The lack of something to support is reason to not pay. Like you, I lack a good local newspaper. There is the nearby city, but they don't cover the important local news well.
Still I'm on the lookout for if someone ever does start covering local news. I don't need national news, even though that is what is cheap easy to get me. (because I get it from to many other places)
There's still valuable info in your local newspaper even if their agenda is the opposite of yours. You just have to read between the lines and phone their bias.
But the biggest stories in your local community will of course be completely hushed many times.
You assume there is one. My 'local' covers high school sports, road construction, but none of the important issues that I need to care about to be an informed voter.
I need someone to be at the library board meeting, school board meeting, city counsel meeting... Some of these happen at the same time so I can't attend them all even if I didn't have other things not mentioned I'm doing then.
It would be cool if you could subscribe to search filters or lists or customizations like you do in uBlock origin, especially community made. Or even sharing rankings or somehow having graphs of useful links. Good manual search is super important because in the age of LLMs it’s the only other option if you want to learn about a topic without relying on ai on the internet, and having good results makes it possible unlike most search engines.
Monero ideally combined with something like this but one or the other would also be an improvement:
>Brave uses a blind token protocol based on Privacy Pass, which decouples payment identity from service usage. These privacy-preserving subscription credentials allow the browser to verify you have a valid purchase of a premium product (like Origin) without learning anything about you.
Depends on how private you want to be, but you can pay Kagi with Bitcoin[1] and use that to buy privacy pass tokens and access it through Tor.[2] It’s not complete privacy, but it decouples the account from the searches.
>I would much rather be able to have things like personalised searches (which are identifying).
This wouldn't affect your accounts usability or features it would just delink your payment from your account. Same as paying for Mullvad with a giftcard bought on Amazon.
Privacy Pass would decouple each search from the account. I think it’s okay to have as an opt-in thing. But private payments should be the default IMO.
I think we’re broadly in agreement: the more private options we have (with different levels of privacy), the better.
It’s possible, in theory, to use Privacy Pass to allow e.g. using credit card to pay for the whole subscription without associating the account with the payment. (E.g. one payment gives you 1 “pass”, and it can be used to extend your account validity for one month.) In practice, it’ll probably be fairly easy to correlate a payment with an account extension, as those would happen almost simultaneously, so a couple more levels are needed here (e.g. doing this through resellers, and maybe breaking up the subscription payment into, say, 30 “day passes” which are used in random order to make it harder to track).
Posteo’s “one-time code system” is a fair bit weaker than Privacy Pass btw: when they get a payment, they can still associate it with your account (so if you’ve paid using, say, bank transfer, they will know your name, and you’ll have to trust them not to store the link). But it can be used together with Privacy Pass :-)
While they don't offer anonymous/private payments, you can use PrivacyPass no matter how you pay for Kagi. So, yes, the fact that you have an account will be known to anyone with access to their subscriber data, but what you search for will not be tied to that identity.
My search engine is a digital product that's already free and private and good and is different to when you give someone your home address to send a package.
I wish there was some kind of EFF-curated "privacy/internet freedom" pass that I could buy.
Kagi is great, but it's not everything. Brave is great, but it's not everything. Signal, GrapheneOS, Tor... this is a battle with many fronts and it's kind of labor intensive to manage a whole bunch of subscriptions/recurring donations.
I'd love to just buy the pass and rest assured that if one of the recipients of that money starts behaving badly, the curator is going to reroute money from them, to whatever the better-behaved alternative is.
The curator (doesn't have to be the EFF, they're just the first to come to mind) can host a https://datatracker.ietf.org/wg/privacypass/about/ token faucet which I use to prove to the various services that I'm among their paid users, but without exposing to any evesdropper any details about myself or that payment. Sometimes such tokens would be unnecessary... it's just a donation... but other times it would be the price of admission. You'd put the coin in the slot to make the thing go. (I'm not talking about cryptocurrency here, it's just a key that deactivates after being used once, and a way to get plenty of them from the curator who is also authoritative about whether you've paid or not).
Paywalls differ by the difficulty of circumventing them - for NYT/WaPo paywalls it's enough to delete cookies for domain, for WSJ/FT you have to use archive.is, and surely there are paywalls with no known workarounds too.
This is the kind of feature that would potentially make me switch, but I feel like at this point AI has disrupted web search for me in such a fundamental way that I rarely use a search engine nowadays.
I'm still getting there, but I've seen this. I spent hours a couple days ago trying to get pi, llamacpp, and qwen3.6 working well. I had them working before, but not well. I Googled, and I followed it's instructions and following the links it found. But everything was a partial answer, and nothing was clear.
I finally decided to ask Claude to do it. It discovered things I didn't even know to look for, immediately making it run twice as fast, and then also helped me do the changes I wanted, like per-model settings.
I knew it was settings.json, but I had it in the wrong spot (thanks, Google) and what I didn't realize is that the 2 settings.json files are different. 1 is global-only, and 1 is per-model-only. If you put the wrong settings in, it'll just ignore them silently. Which left me wondering why it didn't work.
Then it tested my settings. Then I had to use the official webpages to set the proper settings and it tested those, too. It made suggestions about removing or changing some settings that also made things work better and faster.
In less than a hour, it had vastly outstripped everything I'd tried, and it got it right and verified it.
And with how wrong I've seen Google's AI be (and how wrong others have said it is) I can't trust it to even bring me the basics about things. It's worse than useless, it's distracting.
Just be careful with nuance. Someone close to me asks Gemini absolutely everything because they like how it gives you a straight answer, but real life doesn't always work that way. Context is often missing (besides just being wrong at times).
Search engines shouldn't be replaced by AI but they unfortunately are and it result sucks to see.
There are quite a few Luddites like myself that don't trust AI summaries and would rather just have the relevant links, so for us this would be useful. Though TBH I'd rather just stick with the paywalled articles being highlighted as such so I can choose on-the-fly. Perhaps the block option is something I'll turn on when almost all the results are paywalled and I want to see if there is a better option before going to one of them.
Kagi also offers Kagi Assistant, which seems to harness models to apply a search-first behavior, and tends to produce higher-quality results for information that would not normally be within a given model's training data. While the cheaper Kagi plans don't offer the full slate of frontier models, I've found that Kimi models in Kagi Assistant are generally excellent for discovering information, even if I may then go to Claude or Gemini frontier models to parse what I've found.
Kagi Assistant also seems to be a bit lighter on guardrails. I was trying to understand how Libby ebooks are stored and encrypted a few weeks back, and no matter what I did, I could not get any Anthropic products to help me - they insisted I was an evil pirate. Kagi Assistant with either Kimi or Grok (sigh) were happy to help. (If I was doing something nefarious, not sure how far they would have gone, though.)
I believe Kagi has been annotating paywalled results with a circled dollar sign for some time.
This new feature completely weeds such results out of every search.
It would be nice to be able to toggle this (1) in the options drop-down menu on the results page (like Verbatim and Personalized modes), as well as (2) inline (with a "!nopaywall" bang or something).
While this is a nice feature, needing an account (nevermind paying) to then only get a meta search engine will prevent me from ever using it.
I already spread my searches across at least four different engines (Bing, Ecosia, Startpage & Yandex), so no single provider has a complete profile of all my searches.
I don't think your strategy is as effective as you think it is for protecting your privacy online.
You should also read Kagi's privacy policy before implying Kagi would harm your privacy model.
They do offer a mostly anonymous search via the privacy pass. It generates tokens to prove subscription without providing account details. I believe it changes token every session, but it might be every search.
Mostly anonymous because even if they don’t know your account, we all know they can still track your fingerprint.
Keep in mind I have drank the Kagi koolaid, so I’m a bit biased.
I think this is amazing. Love Kagi. I’m happy to pay for a good search.
Nobody speaks about their AI Assistant, but it is really good. They somehow harnessed it so that it mainly searches info first and sticks to the verifiable data. I prefer it to Claude or any other, because it actually answers the question I need without fluff or “Great question! Here’s some plausible nonsense you can trip over instead of doing actual research!”.
You can get it on at least iOS as well as its own app. You can select from bunch of China-models, I use Kim pretty much always. You can also do a system prompt and bunch of the same stuff you can with Claude etc. But it lies to me hardly ever. I don’t know how they did that.
I use pi and librechat and I am really limited by the lack of proper web connector. Testing Tavily right now, Kagi is next.
I wonder why I see so few comments about it on HN, does everybody have this figured out already? Am I a dummy? Am I severely overstating how important of a frature web access and search is for LLM use?
I use Claude code for basically everything, since the harness is pretty complete. The online chat interface works fine too. For basic research I downgrade the model and effort. Smarter models are good at knowing when to web search rather than guess, but you should always specify “do a web search”.
The one caveat is opus has an absolutely awful voice. The way it writes is annoying and verbose. Sonnet actually works better for this type of thing.
You know now that I think of it me too. I absolutely never used the free google one because I never trusted it (you know what they say about free things).
If I'm paying for it through Kagi, I could at least trust their incentives are aligned with mine instead of whoever is paying them the most.
Yeah at first I was very sceptical, I didn’t like their emphasis on AI, and I hate AI summaries generally, but I actually use Kagi AI question answering all the time. I don’t know if it’s just luck or really good UX design but somehow it’s helpful where normally AI features are super annoying.
Part of it might just be that it’s not rammed down your throat anywhere, there’s no sparkle or pop up to “help me write” - google and Microsoft have destroyed their interfaces with ads for AI features. Kagi only does anything if you know what it is and want to use it.
Another +1 for their Asssitant from me. I have the app on my iPhone, and despite having Perplexity/OpenAI/Claude/Gemini paid plans I find myself gravitating to Kagi for q&a.
It truly is a game changer. I often use it before the regular search now. And no joke, it doesn't just save time, I've had it find things that I actually couldn't doing manual search. The reason is because it often knows when there are alternative names for the same sort of product and so can find things that are rebranded or the like.
I agree with you that I'd love to know more about how it works. The citations to real supporting websites are generally very good (though I've seen a few mistakes).
I find it to be hit-and-miss, but overall more useful than not. What I like most is its relative brevity, and the response often helps me find a better prompt to ask so that's good. It's also really fast which I appreciate.
I wouldn't say it's better, it's a great alternative though. I find that for some tasks I much prefer doing a search myself as I may not trust the sources it's pulling from, but for others it's not only adequate but much faster.
I think it's a great complement to human lead search.
"Able to find things that regular search cannot" is pretty much the value-add for AI in Google as well. The blue links are becoming less and less helpful.
I've been a subscriber to Kagi for a few years and just started using the AI assistant.
On the one hand, it's made me a believer in AI as it relates to web searching.
On the other hand, the only reason I need AI to search the web is that the web is now a impenetrable forest of near unreadable, SEO-stuffed, spammy articles.
> On the other hand, the only reason I need AI to search the web is that the web is now a impenetrable forest of near unreadable, SEO-stuffed, spammy articles.
Unfortunately, that is nothing new. A big part of the reason why Google gained popularity, and maintained it for so long, was its resistance to SEO stuffed, spammy articles. The spammers continued escalating things, and now it pretty much takes AI to avoid their garbage.
I remember people using Yahoo and AltaVista a lot. And before gmail everyone had a yahoo email (if they didn't already have a hotmail). Times change, kings fall from grace.
Google has been the dominant force for about 80% of the internet’s existence. I understand things come and go, but Google is incredibly unique. Their importance cannot be overstated
Yes but they were mainly a force of evil. Despite their old motto. They made the internet everything it is today, a hellscape of tracking, advertising and other monetisation.
The new part is AI makes it much easier and cheaper to produce garbage than ever before. So the ratio of garbage to non-garbage has been increasing rapidly.
> So the ratio of garbage to non-garbage has been increasing rapidly.
Is it actually, really? Even late 90s most stuff was garbage, and been like that for as long as I've been on the internet, or maybe I've just always been jaded. But the ratio of "garbage to non-garbage" seems constant for as long as I can remember, multiple decades at this point. The tools for finding the signal through the noise have changed over time though.
I'm not sure AI fully dodges it either. I've asked questions like "best [cuisine] in [place]" and the AI answers with a truly mediocre restaurant that has an amazing website.
Probably true. That said, I spend most of my time querying LLMs for more objective and less financially motivated information. The only time I ask for recommendations is for new routes to explore by bike, and even then it will adapt if you tell it why a route is bad.
I suspect that product and service recommendations will be a near impossible problem to solve. Any time there is a potential for financial gain, someone will figure out how to game the system. Even if it could be resolved, the queries would have to focus on objective criteria ("best" is too subjective).
But then once you get shit results, don't you iterate on the prompt and try again? In this case, add what you actually care about, not just "best cuisine" which can mean a lot of things, and also straight up tell it to not focus on the website or whatever, then eventually you'll get somewhat reasonable and applicable results.
Nope--I was on vacation, walked to the place, and saw it looked like a disaster, but then everyone was hungry and we were already out and about. Too late for more research. I picked a place the old-fashioned way, by walking around and seeing what was nearby. We did fine.
Funny how that works out, because AI is the solution to a problem significantly worsened by AI. The web didn't used to be this terrible before chatGPT, now it's just so much slop. At this point AI is required to sift through all the bullshit caused at least in part by AI.
There is a lot that agents can help with wrt search, but they also have a lot of blind spots. They work well many times, but then there are times where I find stuff that multiple llms missed. Imo this is gonna work as in many other domains, as in the more different tools one is proficient with (searching with llms, searching traditionally etc) the more successfully one solves problems.
I've configured Claude to use their search MCP instead of vanilla Websearch and it has been great. I think it's subscriber-only but I've been on the $5/mo plan for a while and have never him my limit. I might bump up to a higher plan just out of how philosophically refreshing they are as a service. https://github.com/kagisearch/kagimcp
I also confirm the $5 you get is only promotional credit.
I tested on my local LLM chat via Open WebUI, 2 requests from that cost $0.02 (so $4.98 left), and after my monthly Kagi renewal an hour ago, it did not reset, I am still on $4.98.
You are probably better off using Kagi Assistant anyways, as you get $10 there on the $10 Professional plan.
For local LLM web searches, to save on cost I will stick with the Brave Search API which gives you $5 free every month.
> > "Nobody speaks about their AI Assistant, but it is really good. "
So this has nothing to do with the Kagi AI assistant then... it's just the model that's perceived as "good". My point is, there are cheaper ways to get access to the model, without a Kagi subscription...
Kimi K2.7 Code, for me. It's the first model I've used that usually writes working code, somewhat rarely breaks stuff when adding or changing features, and usually fixes a bug correctly just by providing a traceback or description.
Kagi Assistant defaults to the last-used model unless you got there from a Quick Answer. Several times I forgot to change models, and interrupted K2.7 Code, but after a failing to notice the model a few times and getting usable results, I now usually just leave it alone. Unless I want a quicker answer and doubt a "Code" model will grok me.
I remember at the time they were working on it a lot of people (honestly, including me) were pretty critical of them "wasting" their limited resources on AI stuff. We saw it as being sort of similar to the t-shirt drama [1].
I'm here to admit I was wrong, it's incredibly useful. It misses the mark frequently, like any LLM, but it's well annotated with footnotes that link to the original material.
I use it all the time.
I also love how easy it is to use or not - just add a question mark to the end of your query to use the AI, or leave it off for traditional search.
With that UX it doesn't feel like it's being shoved down my throat.
Do you even know what it is? Kagi Assistant is just one part of their offerings. They are a search engine first and foremost. And as you can see many people in this thread are singing the praises of all parts of the Kagi offering.
Creating an account just to hate on a company trying to shake up a monolithic and user-predatory product space is just sad.
Kagi's translation is absolutely top notch. I used it for a few things and while Google's, even Gemini, got some context, kagi's caught the spirit of the text perfectly.
It's that good because the language model they use is expensive I'm pretty sure. I haven't seen many details on what they use, but it's apparently a loss leader for them and already can no longer be used for free.
Meanwhile google translate is probably served on the cheapest model they could muster.
Yes, I use AI assistant a lot. I copy the agent definition from Geir Isene to a custom assistant, it works nicely. In my experience, Kimi K2.6 reasoning works best.
Last time I compared them, it was actually ChatGPT that provided a similarly good search experience, especially compared to Claude. OpenAI doesn't meter its regular chat experience by token usage (yet), so I don't have to ration the better model and just leave it on Thinking High. The end result has definitely been better than what I got from the same queries in Claude or Gemini, and I suspect this isn't just about differences between the models, but also about the way search is integrated into their end-user chat apps.
This is honestly also one of the reasons why I don't subscribe to Kagi anymore. My 'Google-Fu', combined with whatever personalization is happening on my account, already keeps Google's search results in a similarly useful state to Kagi's, so Google and ChatGPT already cover my needs.
Paying for Kagi Ultimate feels like a privacy upgrade but a quality sidegrade that isn't worth $270 a year.
I became so annoyed by google's ai overview and front page results that I am using 99% duckduckgo as the least bad free alternative for the other leg from my chatgpt usage in this context.
I have Kagi and ChatGPT subs and would agree ChatGPT is the better offering. But you can use Luna with Kagi Assistant on the cheap plan which works pretty well too
Is it actually good, like 2006 good? Can I go to page 300 rabbit hole and find some obscure website nobody knows with interesting content on the topic I am researching?
Unfortunately, those days might be good no matter what - a lot of those sites have disappeared with their hosting platforms (e.g Geocities). That said, I have been able to find them on Kagi with a bit of effort.
Kagi also has a Small Web feature which is meant to surface exactly that sort of thing. It's not amazing, but I appreciate it existing. I think you can get a "random small web page" kind of thing even without a subscription here: https://kagi.com/smallweb/
I probably put !expert in half my searches and rely on Kagi for most of my research that isn't just agent context gathering (and a fair amount of what _is_ just context gathering). It's excellent and cites good sources.
I just wish it were easier to keep up with which models are good for what. Especially with the introduction of the research assistant, I have no idea whether I should be using that or one of the standalone models. And don't even get me started on trying to pick the best open model, it's always fluctuating.
I often try 2 different models with the same prompt in case one misses anything. But most of the times I do not see much difference. Newer models do not seem to improve search that much, so I do not think you have to stress too much about new models etc. In general I stick with kimi 2.6 reasoning because it is very very cheap and does thorough searching. I do not see much gains with using more expensive models like opus etc, so I prefer to use multiple cheaper, open models simultaneously except for "harder" queries. The advantage of the cheaper open models is that you can prompt 2-3 of them and if you see the results unsatisfactory you can retry with more expensive ones. And ime sometimes models that are not talked much, like gemini pro, perform quite good.
I think there is a lot of space for improvement on "searching skills" though, sometimes the search strings they use are not great.
I was subscribed to Kagi for a while but cancelled because there was no subscription for JUST search that didn't include their inferior AI assistant. Why am I paying for it?
If you compare the Professional plan (unlimited searches), to the API, you could justify that the extra AI cost is negligable if you are heavy user. The API costs $12 per 1000 searches, so 1.2¢ per individual search. 10/0.0012≈833.33, so at 834 searches or more you would not be paying for the AI features. Obviously only compared to the API.
But if you're using the starter plan, which is 300 searches for $5, then you definitely do pay for the assistant. The API cost would be $3.60.
> They somehow harnessed it so that it mainly searches info first and sticks to the verifiable data.
I like this too, but it isn't as bulletproof as it sounds. I've run into two instances in the past week where it confidently gave me wrong information, because it ran into a search result that had wrong information, and it trusted that information.
You could perhaps say that's fine; if you were doing the search yourself, you'd get the same results and might trust the wrong information too. But the "might" there is... oh boy, I'm gonna say it... load-bearing: someone is much more likely to believe the confident LLM response than if they were looking directly at the primary source, along with search results with other sources that might contradict it.
I use Brave Search+ DuckDuckGo+ Google Image search and that seems to sastify my search requirements while keeping my payment info private.
I don't think I will ever think of subscribing to Kagi metasearch engine, especially since their paywalled search offerings seem to be inferior to other options.
The one thing I find slightly grating about links to Kagi blogs is the top comments are almost always "I use Kagi and it's great!" rather than about the content of the blog. And I'm a happy Kagi subscriber!
I get that this is an option, but what it really shows to me is how broken the model for journalism is. In my view you almost always only get good-quality journalism if you pay for it. (Somehow - e.g. the BBC is paid for by the licence fee, although I'm a bit despairing at the quality of the BBC these days, but perhaps that tracks the significant cuts to their budget...) I'm pretty sure that there's a technically viable solution to micropayments, but there are too many competing interests for anyone to settle on anything. We can't even agree on a single setting that says you don't want advertising cookies!
I know what you mean, but it comes across - even to me, a paying customer who does really like Kagi - a bit like talking to someone involved in an MLM who can't wait to steer the conversation around to their product. I'd just prefer discussions to stay somewhat on topic. There's something interesting in this announcement in that, in my view, it deliberately omits an attribute that's a mild proxy for high-quality journalism, and I always find it a shame when the topic at hand is drowned out by generic boosterism.
> The one thing I find slightly grating about links to Kagi blogs is the top comments are almost always "I use Kagi and it's great!" rather than about the content of the blog. And I'm a happy Kagi subscriber!
Okay, new topic then. Have you heard of InstaPot? ;)
>you almost always only get good-quality journalism if you pay for it
Often true, but sadly doesn't work in reverse - if you pay for it, it often still isn't good-quality journalism (although your mind will tell you it is, just because you paid for it).
> I'm pretty sure that there's a technically viable solution to micropayments, but there are too many competing interests for anyone to settle on anything.
If I could easily use ApplePay to pay 20¢ to read an article, without any tracking or other strings attached, I would probably do that to avoid the whole song and dance of archive workarounds. It’s those competing interests that really get in the way. Giving an email, signing up for an account, storing payment information… money aside, those costs are too high for what I’m getting from a site I don’t read regularly.
It has been tried e.g. https://en.wikipedia.org/wiki/Blendle but it just doesn't work, even a tiny fee will already deter most people from "buying" an article as it will always feel as a loss. Whereas with a subscription you don't feel bad in the moment for losing money but feel good for using your already paid subscription.
Apple news manages to give access to a variety of mainstream sources for a fixed price. Unfortunately there is no way to block the ads and without an ad-free tier it's a no go for me.
> only get good-quality journalism if you pay for it
True for independent media but for corporate media the same stories end up in so many places and most paywalls can be easily removed anyway you aren't missing anything. It also feels wrong to pay for mainstream/corporate media in general, why pay for your own psyop, some PR/AD/NGO/lobby/agency/government already paid for the journalism why should I have to pay to be manipulated? I rather give my money to Democracy Now or Drop Site News.
The biggest problem with micropayments is how credit card payment processors work. You typically pay a percentage fee. But they thought of everything, so you also pay a flat fee per transaction like 10 cents. So they can get their money from tiny transaction.
So for something like a 20 cent micropayment, you pay more to your processor than you pocket.
One solution is a universal token system or something like it. You deposit say, 40 dollars, then get 10,000 tokens or whatever. Like Java Beans from back in the day.
Each card transaction comes with the ability for the buyer to do a chargeback. How would you price this feature, if not per transaction?
The solution here is a government operated utility that works like cash to reduce transaction costs, by not having the protections a card transaction has.
The only way that will work is if each publisher has their own token system, otherwise you have created a new middleman who will want to take their own cut. That middleman will also have to deal with similar problems as payment processors, such as fraud. Perhaps it will be to a lesser degree, because it would be a less valuable target, but it would still add to costs.
Then you run into all sorts of other problems. How do you price articles? I remember running into academic publishers charging the same for a 1 page editorial as a legitimate scientific paper. And how do you charge for the article? Is it 1 token per view or 1 token for unlimited views? If it is the latter, how do you ensure the buyer doesn't lose access? On top of all of that, how do you avoid incentivizing the wrong things? A 1 token article may be 10 times more popular than a 5 token article and cost 1/10th to write, but adjust the price of the 5 token article to 10 tokens and you may cut the readership in half. On top of all of that, the consumer probably doesn't want the cognitive burden of keeping track of all of this stuff.
But the biggest reason why it won't work is because publishers don't want it to work. They have demonstrated time and time again that they would rather sell monthly subscriptions than access to articles.
Of course they'll take their own cut, but that's fine. We accept (sometimes grudgingly) that credit card issuers and networks get their own cut, so there's easy precedent here.
> How do you price articles?
The same way anything else in life is priced: you price at whatever the market can bear.
> And how do you charge for the article? Is it 1 token per view or 1 token for unlimited views?
Also precedent there: online video rental/"purchase". You can decide what model you like better, and see if people want to pay what you ask. You can even offer both, just as is the case for many videos on Amazon, for example.
> but adjust the price of the 5 token article to 10 tokens and you may cut the readership in half.
Yes, that's one of the pitfalls of setting pricing, that every business has to deal with. Note that this problem is present with the monthly subscription model too: half of New York Times subscribers might cancel if the subscription price doubled.
That's a general problem with pricing: determining how much you can raise your price without losing an equivalent (or greater) amount of revenue due to losing customers.
(As an aside, sometimes doubling your prices and exactly halving your customer base is desirable. Fewer customers to deal with at the same overall revenue level might be a net positive. But of course losing a single customer then becomes twice as bad as losing a single customer at the previous pricing.)
> On top of all of that, the consumer probably doesn't want the cognitive burden of keeping track of all of this stuff.
This is the real issue from the consumer side. People like all-you-can-eat plans. They would much rather make a payment decision once, and then get "free" articles to read on an ongoing basis, than have to make a payment decision for every single article they think they might want to read.
> They have demonstrated time and time again that they would rather sell monthly subscriptions than access to articles.
Yup. Recurring subscription revenue is stable and predictable; one-time per-article revenue is not.
> Of course they'll take their own cut, but that's fine. We accept (sometimes grudgingly) that credit card issuers and networks get their own cut, so there's easy precedent here.
In that case, I fail to understand the benefits of using a token system over a credit card. The publisher is loosing a cut to another third party. The consumer is dealing with an extra layer of complexity, and paying more to offset the cost to the publisher.
> This is the real issue from the consumer side. People like all-you-can-eat plans. They would much rather make a payment decision once, and then get "free" articles to read on an ongoing basis, than have to make a payment decision for every single article they think they might want to read.
Do you blame the consumer? Not only do they have to decide whether the article is worth reading, they have to do so sight unseen. While there are some who simply want all-you-can-eat plans, there are plenty who would be happy to pay by the article if they knew the article was digestable (at least in cases where it would be cheaper to buy one article they are interested in, rather than for a month at the buffet table).
It's a real shame the grifters took over cryptocurrencies. I'd love to have a pre-minted ZKP USD style coin. Uncle Sam can take a small transaction fee as consumption tax, people can get wallets as they please. Fully private, semi-decentralized (you don't need full for a government currency), and gov even gets to "tax" transactions on illegal goods (Al Capone). But the two major flaws are that kinda kills a lot of the CC company businesses and it's not great for a country that wants to spy on its citizens. On the flip side, my nightmare is they implement that same system without the ZKP and now your grandma giving you lunch money gets tracked. Weirdly we want with grifting instead. Weird move...
Everyone knows about Monero. If someone doesn't we can count on someone ensuring we do. But I can say with certainty that anyone who knows what a ZKP is knows about Monero even though Monero doesn't use ZKPs
Stablecoins are a pretty clean solution to this. Negligible fee, sub-cent transactions supported, no need for kyc when paying. Fill up your browser periodically, and pay from an embedded wallet.
Whatever the solution is they'd better get to it because I'm not going to subscribe to every site that wants 20/month just because it shows once on HN or in search results.
Your comment illustrates the tension very well, I think. Businesses don't generally want customers who aren't their core market; those people though, often want one slice of a business's offerings. This can lead to big distractions for businesses, as putting in the effort to serve the wrong market is a waste of time. The publishing industry really fell for this when the internet became widespread; they dreamed of huge audiences and big ad numbers. Turns out, that was not a model that worked, at all. The model that works is focusing on serving a niche, and charging a subscription fee, so you can get predictable revenue.
Lots of smart people have tried to implement micropayments like this, and it has never worked.
Content owners would rather have the chance at $10 than give up the chance at $10 for 10c. And customers don’t want to be thinking about their credit balance every time they read an article.
That's not how a micropayments system would work, and you don't need a token system.
It would work similarly to how my transit card works. I use a credit card to put some amount of money on the card itself, say $20, and then every time I use transit, the payment system just deducts $2.85 from my balance. Once my balance goes below a threshold I set, it charges my card again to top up the balance.
So instead of seven $2.85 charges, there's one $20 charge.
A micropayments system would charge you that $20 once, and then deduct 20¢ from your balance every time you read an article. For 100 articles, you've charged the card only once.
It would work similarly on the website side: the micropayments platform wouldn't send a deposit over to the website owner every single time a visitor paid 20¢; it would aggregate payments and either pay out on a timed schedule, or when some minimum dollar amount is reached.
You don't need a token system for this; denominating the "wallet" balance in dollars (or whatever the user's currency is) is fine. A token system might be useful in order to make pricing easier from the website operator's side, but IMO they suck because they hide the actual amount of money the user is paying. It sort of like when you visit a foreign country where you're not familiar with the currency, and so you see that something costs 1,000,000 foobits, and it has no meaning to you, because the exchange rate with your own currency is something like 1:4285.
Used to be you would trade a yak or a horse for something. Then you could trade little pieces of metal. In the future, why not something smaller? I mean, people are trading virtual things (like an app). Why not something small for something as small as a website view?
I'd argue it's still a product issue, one that will eventually be solved.
The reason a payment is too much friction today is because it's surrounded by a raft of cruft: is there an extra "processing fee", am I signing up for a recurring charge and didn't realize it, I have to enter my card details and address, or I have to log in to some system, I have to OTP with my bank, etc.
It hasn't been solved yet, but someone will solve it to the point that it's a single swipe to buy an article for 25c, and it'll work. I would argue the NYT is in the best position to do so, but they are addicted to their scummy business practices where it's hard to cancel, you pay "$2/mo introductory rate" that suddenly escalates to $40/mo, and so on.
The foregone conclusion is that journalism is somehow helpful or valuable to me. I have never read a piece of journalism or "news" and thought to myself, "Wow, how useful! I'm so glad I consumed that."
And I sat here struggling for 5 minutes to think of a single instance. I couldn't come up with one.
So maybe the underlying problem is that journalism isn't nearly as valuable as it thinks it should be.
> We can't even agree on a single setting that says you don't want advertising cookies!
We can't, because the website owners don't want one. They want there to be no standard, because then it's more likely that some people will accidentally allow them.
> The one thing I find slightly grating about links to Kagi blogs is the top comments are almost always "I use Kagi and it's great!"
Couldn't agree more. I don't know how others perceive it, but I find myself becoming more and more wary of broad endorsement. Maybe it's the amount of astroturfing you see on the internet, but I can't help imagine if there is a concerted effort behind it. When someone shares something meaningful and insightful about a product, there's no better endorsement of it. But low effort platitudes are probably doing more harm than intended.
I use kagi. The results quality is pretty good, but I probably use the uprank and downrank feature all the time. Being able to report stuff as AI and get little indicators that a site is AI slop is good.
As an example, serious eats is my go to recipe site, and I've got it tuned so it returns higher in my results.
I also really appreciate I can write my own custom css, and style every box and widget on the page.
And marine most of all, I love that their AI tool will not show up unless you end your query with a "?". Meaning I'm explicitly opting in to using it.
My last reason for liking it is ideological. No ads, no manipulation because of advertising tech. Just ... the stuff I searched for.
I do not like the image search page, however. It feels clunky in a way I can't put my finger on.
There's a few other things I really like, but it's hard to type them all out. So I often go, "kagi is great, love kagi."
The more I think about it, the less sure I am that the model for journalism is actually broken. For some unknown reason, people do not want to subscribe to news media, but they'll happily pay for Netflix, Disney, HBO and Spotify.
The comment is always: I want to be able to pay for this one article, but streaming doesn't work like that. I can watch a movie on Netflix and pay €25 per month, then watch what ever else I want. Or, I can pay €13 and watch just that one movie on AppleTV and nothing else. Newspapers are more expensive, typically, because their subscriber count is way lower than that of Netflix, but their cost remain roughly the same.
You can't say: I want to pay $1 for an article, unless that article is massively popular, that's not going to cover all the other less popular, but important news for that day. Journalism requires a stable income to produce things of value. Investigations may that months or years. Using the cost of the Netflix/AppleTV example you should either subscribe, or pay around half the monthly subscription per article.
So no, the revenue model for journalism isn't broken. Our ability so recognize the value of good journalism is. It's easier and more profitable to write an "article" about Kim Kardashians oatmeal with a few dusin ads slapped on, than it is to do a multiple page article about the governance problems of the UK, which you need to sell, because advertisers don't want to be associated with politics and the target audience big enough anyway.
So, we can't have good, independent journalism, because the amount of people who care aren't enough to cover the expense or they want to shop around and only pay for the topics and views they care about and agree with. Micro-payments is the complete wrong way to go, we need more people to start subscribing to newspapers again and the papers needs to take that money and funnel that money directly into actual journalism.
I wish Kagi had existed years ago. I just saw someone talking about the Google SEO-laden experience when troubleshooting coding errors and realized it’s a pain I haven’t felt in a long time. It’s telling that Google never made a “remove this site from results” option. Their incentives weren’t set up to provide the best user experience.
289 comments
[ 5.2 ms ] story [ 486 ms ] thread[1] https://en.wikipedia.org/wiki/Bypass_Paywalls_Clean
Even as LLMs slurp up most of the internet and replace search I think Kagi is still useful.
Reddit has recently blocked access to old reddit without an account and the ability to filter out stuff like that is useful.
Friends send me links to reddit and stuff so it's handy to have an automatic redirect.
Edit: Actually I think this was part of their documentation to explain the intended use of that feature. Too bad this will now no longer work!
(https://help.kagi.com/kagi/features/redirects.html, see example 1)
I don't understand why this is a barrier considering their accounts aren't invasive at all since you can use a temporary email account to sign up, and then you never have to do anything via email after that.
I created an account recently after not having one for a couple years. I was banned within 2 weeks. I accidentally connected to my VPN when I had a Reddit page open in a minimized window that I forgot about. I assume that’s what did it. Last I checked, I couldn’t even login with this account anymore.
If that doesn't bother you, then good for you, and we have different feelings about privacy.
I'm not logging in to my private account from a work machine that's MITMd for "security", just to browse search results or take a short break!
(I've been a user for nearly 21 years, but using it less and less with each stupid choice - these days I post a comment maybe once a week there. I think I may finally be done with Reddit, sadly. It was good while it lasted.)
That'd be an important milestone in enshittification, when it comes to other entities that were happy to outsource their fan-forums, support, announcement systems, etc.
Kind of like how many groups started using Twitter as a way to provide a stream of announcements--sometimes time-critical ones--to a broad audience. The entire use-case goes down the toilet when users aren't being shown your stuff in chronological order anymore.
This is really it though isn't it
I know new reddit has its potential inspiration in stopping LLM scraping(?) but websites are made worse because of some stupid metric that has to be fulfilled
Cool.
That's why it is a choice for companies to not pay for (human) software engineers anymore for writing software and to get all the agents to do the work and write the software that we use today.
Just like how we will see articles now being written by AI that will rank in Kagi instead of human written articles.
If a news site wants to join the ranks of my RSS feeds they need to be of high enough quality that their reputation does more work than paywalled search results do.
Nothing worse than clicking on it, and after 5 seconds you get the naggy pop-up that disables the scrolling. Now I need to send it to archive.ph to read it.
I'm fine with a paywall, but put it above, don't interrupt me as I am reading.
I think they have no imagination - like most rich kid “founders” they have no original ideas.
Would never use this trash
That's why this is a good feature.
I also don't care if people have original ideas if they take an existing idea and do it better, which is what I appreciate about Kagi at the moment. I don't want anything interesting or novel. I want search that doesn't berate me with ads or AI overviews if I don't ask for them.
However tracking that is tricky at best.
Anyway, you can block sources individually in Kagi, that can weed out the most common paywall options you don't have, like Forbes or whatever.
Still I'm on the lookout for if someone ever does start covering local news. I don't need national news, even though that is what is cheap easy to get me. (because I get it from to many other places)
But the biggest stories in your local community will of course be completely hushed many times.
I need someone to be at the library board meeting, school board meeting, city counsel meeting... Some of these happen at the same time so I can't attend them all even if I didn't have other things not mentioned I'm doing then.
*binding arbitration agreement required
>Brave uses a blind token protocol based on Privacy Pass, which decouples payment identity from service usage. These privacy-preserving subscription credentials allow the browser to verify you have a valid purchase of a premium product (like Origin) without learning anything about you.
https://brave.com/origin/
[1] https://help.kagi.com/kagi/plans/payment-methods.html
[2] https://help.kagi.com/kagi/privacy/tor.html
I would much rather be able to have things like personalised searches (which are identifying).
This wouldn't affect your accounts usability or features it would just delink your payment from your account. Same as paying for Mullvad with a giftcard bought on Amazon.
I don't really understand your first paragraph.
It’s possible, in theory, to use Privacy Pass to allow e.g. using credit card to pay for the whole subscription without associating the account with the payment. (E.g. one payment gives you 1 “pass”, and it can be used to extend your account validity for one month.) In practice, it’ll probably be fairly easy to correlate a payment with an account extension, as those would happen almost simultaneously, so a couple more levels are needed here (e.g. doing this through resellers, and maybe breaking up the subscription payment into, say, 30 “day passes” which are used in random order to make it harder to track).
Posteo’s “one-time code system” is a fair bit weaker than Privacy Pass btw: when they get a payment, they can still associate it with your account (so if you’ve paid using, say, bank transfer, they will know your name, and you’ll have to trust them not to store the link). But it can be used together with Privacy Pass :-)
The results are quite good except images I like the results more then DDG or Startpage.
Kagi is great, but it's not everything. Brave is great, but it's not everything. Signal, GrapheneOS, Tor... this is a battle with many fronts and it's kind of labor intensive to manage a whole bunch of subscriptions/recurring donations.
I'd love to just buy the pass and rest assured that if one of the recipients of that money starts behaving badly, the curator is going to reroute money from them, to whatever the better-behaved alternative is.
The curator (doesn't have to be the EFF, they're just the first to come to mind) can host a https://datatracker.ietf.org/wg/privacypass/about/ token faucet which I use to prove to the various services that I'm among their paid users, but without exposing to any evesdropper any details about myself or that payment. Sometimes such tokens would be unnecessary... it's just a donation... but other times it would be the price of admission. You'd put the coin in the slot to make the thing go. (I'm not talking about cryptocurrency here, it's just a key that deactivates after being used once, and a way to get plenty of them from the curator who is also authoritative about whether you've paid or not).
I finally decided to ask Claude to do it. It discovered things I didn't even know to look for, immediately making it run twice as fast, and then also helped me do the changes I wanted, like per-model settings.
I knew it was settings.json, but I had it in the wrong spot (thanks, Google) and what I didn't realize is that the 2 settings.json files are different. 1 is global-only, and 1 is per-model-only. If you put the wrong settings in, it'll just ignore them silently. Which left me wondering why it didn't work.
Then it tested my settings. Then I had to use the official webpages to set the proper settings and it tested those, too. It made suggestions about removing or changing some settings that also made things work better and faster.
In less than a hour, it had vastly outstripped everything I'd tried, and it got it right and verified it.
And with how wrong I've seen Google's AI be (and how wrong others have said it is) I can't trust it to even bring me the basics about things. It's worse than useless, it's distracting.
Search engines shouldn't be replaced by AI but they unfortunately are and it result sucks to see.
Kagi Assistant also seems to be a bit lighter on guardrails. I was trying to understand how Libby ebooks are stored and encrypted a few weeks back, and no matter what I did, I could not get any Anthropic products to help me - they insisted I was an evil pirate. Kagi Assistant with either Kimi or Grok (sigh) were happy to help. (If I was doing something nefarious, not sure how far they would have gone, though.)
This new feature completely weeds such results out of every search.
It would be nice to be able to toggle this (1) in the options drop-down menu on the results page (like Verbatim and Personalized modes), as well as (2) inline (with a "!nopaywall" bang or something).
I already spread my searches across at least four different engines (Bing, Ecosia, Startpage & Yandex), so no single provider has a complete profile of all my searches.
I certainly won't be paying to hand over my data.
Mostly anonymous because even if they don’t know your account, we all know they can still track your fingerprint.
Keep in mind I have drank the Kagi koolaid, so I’m a bit biased.
Nobody speaks about their AI Assistant, but it is really good. They somehow harnessed it so that it mainly searches info first and sticks to the verifiable data. I prefer it to Claude or any other, because it actually answers the question I need without fluff or “Great question! Here’s some plausible nonsense you can trip over instead of doing actual research!”.
You can get it on at least iOS as well as its own app. You can select from bunch of China-models, I use Kim pretty much always. You can also do a system prompt and bunch of the same stuff you can with Claude etc. But it lies to me hardly ever. I don’t know how they did that.
If only Kagi had a Shopping feature, then it would be a complete solution for us.
Why?
But, regular models and harnesses work okay for this too. You just have to specify “web search”, very important.
I use pi and librechat and I am really limited by the lack of proper web connector. Testing Tavily right now, Kagi is next.
I wonder why I see so few comments about it on HN, does everybody have this figured out already? Am I a dummy? Am I severely overstating how important of a frature web access and search is for LLM use?
The one caveat is opus has an absolutely awful voice. The way it writes is annoying and verbose. Sonnet actually works better for this type of thing.
If I'm paying for it through Kagi, I could at least trust their incentives are aligned with mine instead of whoever is paying them the most.
Part of it might just be that it’s not rammed down your throat anywhere, there’s no sparkle or pop up to “help me write” - google and Microsoft have destroyed their interfaces with ads for AI features. Kagi only does anything if you know what it is and want to use it.
I agree with you that I'd love to know more about how it works. The citations to real supporting websites are generally very good (though I've seen a few mistakes).
I think it's a great complement to human lead search.
On the one hand, it's made me a believer in AI as it relates to web searching.
On the other hand, the only reason I need AI to search the web is that the web is now a impenetrable forest of near unreadable, SEO-stuffed, spammy articles.
Unfortunately, that is nothing new. A big part of the reason why Google gained popularity, and maintained it for so long, was its resistance to SEO stuffed, spammy articles. The spammers continued escalating things, and now it pretty much takes AI to avoid their garbage.
It doesn’t change the fact that they literally impacted the trajectory of how the internet developed in a very obvious, concrete way way.
Is it actually, really? Even late 90s most stuff was garbage, and been like that for as long as I've been on the internet, or maybe I've just always been jaded. But the ratio of "garbage to non-garbage" seems constant for as long as I can remember, multiple decades at this point. The tools for finding the signal through the noise have changed over time though.
I suspect that product and service recommendations will be a near impossible problem to solve. Any time there is a potential for financial gain, someone will figure out how to game the system. Even if it could be resolved, the queries would have to focus on objective criteria ("best" is too subjective).
ChatGPT is trained on all of Reddit and internet forums where most of the useful info is. It just automates the drudgery.
There is a lot that agents can help with wrt search, but they also have a lot of blind spots. They work well many times, but then there are times where I find stuff that multiple llms missed. Imo this is gonna work as in many other domains, as in the more different tools one is proficient with (searching with llms, searching traditionally etc) the more successfully one solves problems.
EDIT: I see I have $5 of "pre-paid API credit". Is this a one-time thing or is that included with my plan?
Edit: this is a one-time promotional credit. One must pay for credits thereafter, in addition to their subscription.
I tested on my local LLM chat via Open WebUI, 2 requests from that cost $0.02 (so $4.98 left), and after my monthly Kagi renewal an hour ago, it did not reset, I am still on $4.98.
You are probably better off using Kagi Assistant anyways, as you get $10 there on the $10 Professional plan.
For local LLM web searches, to save on cost I will stick with the Brave Search API which gives you $5 free every month.
which model?
So this has nothing to do with the Kagi AI assistant then... it's just the model that's perceived as "good". My point is, there are cheaper ways to get access to the model, without a Kagi subscription...
Kagi Assistant defaults to the last-used model unless you got there from a Quick Answer. Several times I forgot to change models, and interrupted K2.7 Code, but after a failing to notice the model a few times and getting usable results, I now usually just leave it alone. Unless I want a quicker answer and doubt a "Code" model will grok me.
I'm here to admit I was wrong, it's incredibly useful. It misses the mark frequently, like any LLM, but it's well annotated with footnotes that link to the original material.
I use it all the time.
I also love how easy it is to use or not - just add a question mark to the end of your query to use the AI, or leave it off for traditional search.
With that UX it doesn't feel like it's being shoved down my throat.
[1] https://blog.kagi.com/celebrating-20k
Creating an account just to hate on a company trying to shake up a monolithic and user-predatory product space is just sad.
Meanwhile google translate is probably served on the cheapest model they could muster.
https://github.com/isene/ClaudeCode/blob/main/agents/web-res...
This is honestly also one of the reasons why I don't subscribe to Kagi anymore. My 'Google-Fu', combined with whatever personalization is happening on my account, already keeps Google's search results in a similarly useful state to Kagi's, so Google and ChatGPT already cover my needs.
Paying for Kagi Ultimate feels like a privacy upgrade but a quality sidegrade that isn't worth $270 a year.
Kagi also has a Small Web feature which is meant to surface exactly that sort of thing. It's not amazing, but I appreciate it existing. I think you can get a "random small web page" kind of thing even without a subscription here: https://kagi.com/smallweb/
I think there is a lot of space for improvement on "searching skills" though, sometimes the search strings they use are not great.
But if you're using the starter plan, which is 300 searches for $5, then you definitely do pay for the assistant. The API cost would be $3.60.
I like this too, but it isn't as bulletproof as it sounds. I've run into two instances in the past week where it confidently gave me wrong information, because it ran into a search result that had wrong information, and it trusted that information.
You could perhaps say that's fine; if you were doing the search yourself, you'd get the same results and might trust the wrong information too. But the "might" there is... oh boy, I'm gonna say it... load-bearing: someone is much more likely to believe the confident LLM response than if they were looking directly at the primary source, along with search results with other sources that might contradict it.
The assistant is still good and I also use it regularly. It’s also nice to not have to leave the terminal sometimes.
https://github.com/themauveavenger/pi-kagi
I don't think I will ever think of subscribing to Kagi metasearch engine, especially since their paywalled search offerings seem to be inferior to other options.
Right up there with Bluesky.
Pointless shit that nobody will ever use.
When will this be available into HN, too ?
hn already disallows paywalled links, unless there is an easy/low-effort bypass (noscript, archive link, gift link, etc.).
I get that this is an option, but what it really shows to me is how broken the model for journalism is. In my view you almost always only get good-quality journalism if you pay for it. (Somehow - e.g. the BBC is paid for by the licence fee, although I'm a bit despairing at the quality of the BBC these days, but perhaps that tracks the significant cuts to their budget...) I'm pretty sure that there's a technically viable solution to micropayments, but there are too many competing interests for anyone to settle on anything. We can't even agree on a single setting that says you don't want advertising cookies!
Okay, new topic then. Have you heard of InstaPot? ;)
It is one of the very, very few digital services that hasn't fucked me over within a year.
I will celebrate as if I had discovered Kagi yesterday.
Often true, but sadly doesn't work in reverse - if you pay for it, it often still isn't good-quality journalism (although your mind will tell you it is, just because you paid for it).
If I could easily use ApplePay to pay 20¢ to read an article, without any tracking or other strings attached, I would probably do that to avoid the whole song and dance of archive workarounds. It’s those competing interests that really get in the way. Giving an email, signing up for an account, storing payment information… money aside, those costs are too high for what I’m getting from a site I don’t read regularly.
True for independent media but for corporate media the same stories end up in so many places and most paywalls can be easily removed anyway you aren't missing anything. It also feels wrong to pay for mainstream/corporate media in general, why pay for your own psyop, some PR/AD/NGO/lobby/agency/government already paid for the journalism why should I have to pay to be manipulated? I rather give my money to Democracy Now or Drop Site News.
So for something like a 20 cent micropayment, you pay more to your processor than you pocket.
One solution is a universal token system or something like it. You deposit say, 40 dollars, then get 10,000 tokens or whatever. Like Java Beans from back in the day.
The solution here is a government operated utility that works like cash to reduce transaction costs, by not having the protections a card transaction has.
Then you run into all sorts of other problems. How do you price articles? I remember running into academic publishers charging the same for a 1 page editorial as a legitimate scientific paper. And how do you charge for the article? Is it 1 token per view or 1 token for unlimited views? If it is the latter, how do you ensure the buyer doesn't lose access? On top of all of that, how do you avoid incentivizing the wrong things? A 1 token article may be 10 times more popular than a 5 token article and cost 1/10th to write, but adjust the price of the 5 token article to 10 tokens and you may cut the readership in half. On top of all of that, the consumer probably doesn't want the cognitive burden of keeping track of all of this stuff.
But the biggest reason why it won't work is because publishers don't want it to work. They have demonstrated time and time again that they would rather sell monthly subscriptions than access to articles.
> How do you price articles?
The same way anything else in life is priced: you price at whatever the market can bear.
> And how do you charge for the article? Is it 1 token per view or 1 token for unlimited views?
Also precedent there: online video rental/"purchase". You can decide what model you like better, and see if people want to pay what you ask. You can even offer both, just as is the case for many videos on Amazon, for example.
> but adjust the price of the 5 token article to 10 tokens and you may cut the readership in half.
Yes, that's one of the pitfalls of setting pricing, that every business has to deal with. Note that this problem is present with the monthly subscription model too: half of New York Times subscribers might cancel if the subscription price doubled.
That's a general problem with pricing: determining how much you can raise your price without losing an equivalent (or greater) amount of revenue due to losing customers.
(As an aside, sometimes doubling your prices and exactly halving your customer base is desirable. Fewer customers to deal with at the same overall revenue level might be a net positive. But of course losing a single customer then becomes twice as bad as losing a single customer at the previous pricing.)
> On top of all of that, the consumer probably doesn't want the cognitive burden of keeping track of all of this stuff.
This is the real issue from the consumer side. People like all-you-can-eat plans. They would much rather make a payment decision once, and then get "free" articles to read on an ongoing basis, than have to make a payment decision for every single article they think they might want to read.
> They have demonstrated time and time again that they would rather sell monthly subscriptions than access to articles.
Yup. Recurring subscription revenue is stable and predictable; one-time per-article revenue is not.
In that case, I fail to understand the benefits of using a token system over a credit card. The publisher is loosing a cut to another third party. The consumer is dealing with an extra layer of complexity, and paying more to offset the cost to the publisher.
> This is the real issue from the consumer side. People like all-you-can-eat plans. They would much rather make a payment decision once, and then get "free" articles to read on an ongoing basis, than have to make a payment decision for every single article they think they might want to read.
Do you blame the consumer? Not only do they have to decide whether the article is worth reading, they have to do so sight unseen. While there are some who simply want all-you-can-eat plans, there are plenty who would be happy to pay by the article if they knew the article was digestable (at least in cases where it would be cheaper to buy one article they are interested in, rather than for a month at the buffet table).
We still have Monero. Here's some recent projects.
https://donatemonero.org/
https://monerostats.org/
https://monero.forum/ notice zero posts on the first page about USD price besides one disparaging Bitcoiners for caring about USD price.
They work very well for LLM tokens though. So, go ahead and get your news that way.
Content owners would rather have the chance at $10 than give up the chance at $10 for 10c. And customers don’t want to be thinking about their credit balance every time they read an article.
It would work similarly to how my transit card works. I use a credit card to put some amount of money on the card itself, say $20, and then every time I use transit, the payment system just deducts $2.85 from my balance. Once my balance goes below a threshold I set, it charges my card again to top up the balance.
So instead of seven $2.85 charges, there's one $20 charge.
A micropayments system would charge you that $20 once, and then deduct 20¢ from your balance every time you read an article. For 100 articles, you've charged the card only once.
It would work similarly on the website side: the micropayments platform wouldn't send a deposit over to the website owner every single time a visitor paid 20¢; it would aggregate payments and either pay out on a timed schedule, or when some minimum dollar amount is reached.
You don't need a token system for this; denominating the "wallet" balance in dollars (or whatever the user's currency is) is fine. A token system might be useful in order to make pricing easier from the website operator's side, but IMO they suck because they hide the actual amount of money the user is paying. It sort of like when you visit a foreign country where you're not familiar with the currency, and so you see that something costs 1,000,000 foobits, and it has no meaning to you, because the exchange rate with your own currency is something like 1:4285.
https://nakamotoinstitute.org/library/micropayments-and-ment...
The reason a payment is too much friction today is because it's surrounded by a raft of cruft: is there an extra "processing fee", am I signing up for a recurring charge and didn't realize it, I have to enter my card details and address, or I have to log in to some system, I have to OTP with my bank, etc.
It hasn't been solved yet, but someone will solve it to the point that it's a single swipe to buy an article for 25c, and it'll work. I would argue the NYT is in the best position to do so, but they are addicted to their scummy business practices where it's hard to cancel, you pay "$2/mo introductory rate" that suddenly escalates to $40/mo, and so on.
I could see Substack taking a crack at it also.
And I sat here struggling for 5 minutes to think of a single instance. I couldn't come up with one.
So maybe the underlying problem is that journalism isn't nearly as valuable as it thinks it should be.
We can't, because the website owners don't want one. They want there to be no standard, because then it's more likely that some people will accidentally allow them.
Couldn't agree more. I don't know how others perceive it, but I find myself becoming more and more wary of broad endorsement. Maybe it's the amount of astroturfing you see on the internet, but I can't help imagine if there is a concerted effort behind it. When someone shares something meaningful and insightful about a product, there's no better endorsement of it. But low effort platitudes are probably doing more harm than intended.
I use kagi. The results quality is pretty good, but I probably use the uprank and downrank feature all the time. Being able to report stuff as AI and get little indicators that a site is AI slop is good.
As an example, serious eats is my go to recipe site, and I've got it tuned so it returns higher in my results.
I also really appreciate I can write my own custom css, and style every box and widget on the page.
And marine most of all, I love that their AI tool will not show up unless you end your query with a "?". Meaning I'm explicitly opting in to using it.
My last reason for liking it is ideological. No ads, no manipulation because of advertising tech. Just ... the stuff I searched for.
I do not like the image search page, however. It feels clunky in a way I can't put my finger on.
There's a few other things I really like, but it's hard to type them all out. So I often go, "kagi is great, love kagi."
- Out of the box, search result quality is comparable to Google
- Search is slow. Latency for most queries is about a second. Sometimes 3 seconds. Google is shockingly fast in comparison.
- Personalized results is a killer feature (blocking, boosting and suppressing domains)
- There is no iOS app. Only one in beta that is seemingly abandoned.
- Their site/app design is just passable at best
- Their AI quick answers give solid answers and are pretty concise
The comment is always: I want to be able to pay for this one article, but streaming doesn't work like that. I can watch a movie on Netflix and pay €25 per month, then watch what ever else I want. Or, I can pay €13 and watch just that one movie on AppleTV and nothing else. Newspapers are more expensive, typically, because their subscriber count is way lower than that of Netflix, but their cost remain roughly the same.
You can't say: I want to pay $1 for an article, unless that article is massively popular, that's not going to cover all the other less popular, but important news for that day. Journalism requires a stable income to produce things of value. Investigations may that months or years. Using the cost of the Netflix/AppleTV example you should either subscribe, or pay around half the monthly subscription per article.
So no, the revenue model for journalism isn't broken. Our ability so recognize the value of good journalism is. It's easier and more profitable to write an "article" about Kim Kardashians oatmeal with a few dusin ads slapped on, than it is to do a multiple page article about the governance problems of the UK, which you need to sell, because advertisers don't want to be associated with politics and the target audience big enough anyway.
So, we can't have good, independent journalism, because the amount of people who care aren't enough to cover the expense or they want to shop around and only pay for the topics and views they care about and agree with. Micro-payments is the complete wrong way to go, we need more people to start subscribing to newspapers again and the papers needs to take that money and funnel that money directly into actual journalism.