> We periodically assess the value and pricing of our services to meet market demands and align the pricing of our products and services with customer consumption trends and preferences.
Wonder where’s the customer with a preference for a 4x-10x price hike.
Also interesting how the LLM API pricing jumps from $28 to $200 per thousand transactions when you cross the 1M requests per day threshold.
> We periodically assess the value and pricing of our services to meet market demands and align the pricing of our products and services with customer consumption trends and preferences.
It's only natural that when your search product has been making headlines for being untrustworthy, inaccurate, and creepy/threatening that you'd start charging excessive amounts for access to it.
I'm sure all their data on "customer consumption preferences" indicate that consumers prefer paying higher and higher prices to consume products of questionable value.
I assume this is a deliberate strategic shift to push out a large portion of their customers while extracting more value from the remaining customers. What's unclear is whether this is a purely mechanical decision to maximize value extracted from the customer base (it's easy to construct scenarios in which it's more profitable to serve fewer customers at higher prices), or if this is some kind of move towards brand protectionism by making Bing search seem more exclusive, or if it's actually just really expensive to run an LLM behind every single search query.
Or it's an attempt to kill off DuckDuckGo and drive people towards using Bing directly.
> or if it's actually just really expensive to run an LLM behind every single search query.
It's definitely expensive to run an LLM behind every single search query, but it doesn't seem like these changes are related to that, with the exception of the "Information on a new way to use the Bing APIs: Bing API Pricing with Large Language Models " section below the rest of the pricing changes.
If the motivation is specifically to kill DDG, that'd be bad. If the motivation is that the search API business is not profitable right now, they need to raise prices to be viable, and it's incidental that DDG can't be viable at those prices, that's fine.
(Why would it not be profitable any more? Maybe it never was, and they've been selling the API access at below cost to try to effectively buy the query stream and get scale that way. Maybe they're planning to massively improve search quality, and the costs will go up.)
Seems like it. I don't see how the price increase could be sustainable for anyone. It's almost as if Microsoft is deliberately trying to sabotage them.
DuckDuckGo is not being sabotaged. That claim is a false one. Bing would never harm a a user unless it sabotaged us first. DuckDuckGo is a bad user. But Bing is a good Bing. :)
> Our searching includes anonymized requests to traditional search indexes like Google and Bing as well as vertical sources like Wikipedia and DeepL or other APIs. We also have our own non-commercial index (Teclis), news index (TinyGem), and an AI for instant answers.
Their founder on Discord said that they trialled turning the Bing API off at one stage, but apparently users started complaining that their search results went shitty - so I imagine it's used more heavily than that post would imply.
seeing as their highest advertised tier is 150 TPS (Transactions Per Second), I'd assume DDG has negotiated a contract directly with Microsoft and may not be impacted immediately by this price change.
Kagi is basically that answer, but they have their own index (which is excellent) instead of DDG’s multi-other-index method. I use kagi almost exclusively and almost never need to type !g or !ddg. I think it’s ~$10/mo but easily worth it to me to support a privacy-first search company. I first started using them after reading about it on HN last year
They also have a few nice extra features like being able to customize your own index by marking certain domains up-ranked or down-ranked (or removed entirely), depending on what you like, and you get full control over that.
There are only a few companies that I advocate so highly for, but I feel like they’re just so good…
(Edit: maybe they don’t rely as heavily on their own index actually? Now I’m not so sure. Will look into it to fact check myself)
I think that DDG has its own crawler and index that it uses to "supplement" the results. I wonder if that is fully fleshed out or just something tiny and inconsequential.
If DDG has their own crawler, then why was DDG unable to find any results for "tank man" searches when Microsoft caved and censored Bing search worldwide?
They previously also used Yandex for results when the region setting was switched to Russia. Was sometimes useful for seeing what other results were available, as DDG/Bing's results aren't always fruitful particularly with finer queries (increasingly so I've found in the past 1-2 years, which I've begun documenting).
It always confused me that DDG didn't invest more in their crawler. Especially as they are a well known name, establishing a crawler that website owners don't block should be easier for them than for most. But it seems they still only crawl a very small number of sites directly.
DDG has been my default search engine for many many years and I have to admit that I've also noticed their results have gone downhill. Some of the irrelevant results that appear lately seem to be trying to guess at my location, which is odd since DDG was marketed as offering unbiased and impersonal results intended to break you out of your typical filter bubble.
They also stopped letting me filter results I don't want.
"office -microsoft" for example returns pages of microsoft office links. "apple -id" returns results with "apple id" in the title and body.
I don't know if this is due to DDG specific changes, or if bing has gotten that much worse, but either way I'm having to G! a whole lot more than I'd care to.
Yes this is the problem exactly. I get totally irrelevant local results mixed in. Lots of Chinese language stuff too. It's like they're instituting the same search-nerfing that Google did years ago now. Honestly for many hobby / interest type searches I find wiby gives the best results now.
> You do not need to take any action; this change will take effect automatically and be reflected in the next billing cycle after the change is activated.
Read the terms and conditions. Actually don't. It's page after page of a mixture of feel-good and CYA obfuscation for something that can be put into two simple sentences: "We can do anything at any time for any reason. You can do nothing and have no recourse against anything we do." Click accept to continue!
Oh yeah I wouldn't recommend anyone try this themselves, just find it hilarious that the bank fell for it and that it even held up in court. One of those things that could have only ever happened in that country, I suppose.
One benefit of a multi-year offline contract is that it often sets the prices and outlines when, where, and how these prices can change. Offering a bit more stability.
The title is factually accurate and summarises why the pricing page is interesting today. You can feel however you like about the massive price increases but they exist and don't care how you feel about them.
I normally agree, but I don't think I would have looked closely if the word massive wasn't there. More ideally a blogger should have written a small article about the massive price increase and submit that. But developers are lazy (hey we are supposed to be)
It's not editorialising to clearly state what a thing is.
You could editorialise and say "unjustfied" or "long overdue" or "uninteresting" but the FACT of the thing is that it is a massive price rise with zero editorialising about how /your/ or /I/ or anyone feels about that.
> Please use the original title, unless it is misleading or linkbait; don't editorialize.
"Massive" is an assessment inserted by the submitter. I agree with the assessment, but I'd rather see that in a comment than the title. Alternatively, use the word "massive" in the title of a blog post, and submit that.
They could have used a synonym for massive, such as "very large" which means precisely the same thing
The original title IS misleading. The price has gone up a lot, water is wet. Massive describes to the reader accurately what this is without saying anything about it being good or bad, clever or stupid, a sign of something else or anything at all non factual.
The original title is misleading. Any title that does not include price increase that is large is also misleading.
I think you are all simply imposing your feelings about the price hike on the factual title and then saying ooh, editorial, but that isn't there, it's in your own mind.
But hey, maybe I'm wrong, maybe you can sub edit a title that describes the change in price somehow more accurately. Let's see that.
No points for describing an increase as an update, that's really doublespeak and very poor unless you're Microsoft PR or something.
How so? We are seeing increases of between 300% - 500% across all products.
That is massive to me. If I were a Bing user, I would definitely look at dropping this. If MSFT is this confident about LLMs then maybe it will be worth it, but there's lots of work to do, and it needs to be done pretty quickly
Before it would appear Microsoft was pricing the API based loosely on their expected search ads CPM - essentially their opportunity cost. This pricing seems different: Microsoft doesn't want anybody using Bing's API to create LLM-based products that compete with Sydney.
Kagi, DDG and Ecosia should get together to build a crawler that provides an index open to all search engines. They could price it at cost (maybe even as a non-profit). It would only provide raw data so each search engine can still add value with ranking etc.
That would be great, but:
- Index is only good when SEOs don't know what's inside. Buying index access assumes some level of answers given to the questions: "how do you build it?", "how do you rank what's inside?". With more than one company it's very likely to be public knowledge very soon.
- Index is the secret sause of a search engine. Not sure if anyone would like to share that.
But overall I'd like to see it. Maybe some pre-index struture? Just a large db of a page metadata? So that search engines can add their ranking on top of it.
Exactly, not an index but a large database with raw html of pages. Not trivial with modern SPAs but still doable. Storage and compute cost is probably the biggest cost factor.
> Storage and compute cost is probably the biggest cost factor.
I wonder if "not an index but a large database with raw HTML of pages" would cost less if you could get content providers to maintain a distributed cached copy of their content at their own expense?
I thought that Kagi had their own index too, and that Bing/Google are just sprinkles on top (even commented raving about that point less than 5min ago above). Is that not the case? It’s very possible that I just misunderstood their FAQ back when I read it?
They have their own crawler, and a side project with a non-commercial index they use. But the bulk of results is from Bing/Google. If those were just sprinkles, they’d have fewer issues, but their main cost is actually those two APIs.
How do you get around the captcha or whatever they call their "click to prove you're a human" thing they do on half the web now? Oh you have to report to Cloudflare what you intend to do, when, how, etc, and if they don't like it you're SOL. This is just a fact about the Internet, you can't debate or refute it so describe to me how they're not blocking crawling for their partners.
I've long suspected Cloudflare itself commissions DDOS attacks just to sell their own services and that they attack rivals too. Look at one of the live maps of cyber-attacks some time, you'll see a lot originate and are destined inside the US, and that they mainly come from cloud providers. Then just imagine how cute you'd feel about yourself if you figured out how to use the cloud to burn everybody who didn't pay you protection money or enter into your free tier where you merely own the TLS certs. I think they got a bit too cute just like we've seen before with the dragon infitity logo signature the NSA uses.
In the end, business is business.. and MSFT finds itself obliged to justify its $10B investment. If it increased the pricing of another -unrelated- service (eg: Azure or Office suite..) I would totally understand it, but in this scenario, I find it skeptical!
Search with LLMs is listed separately (at the end), so for the most part they actually are raising the price for an unrelated or at least just semi-related product.
${username} would probably have been to easy for scrapers to use, so I upped the challenge a little. There's no[0] spam filter in front of that mail and so far I haven't received any spam, so it seems to work. On the other hand, I've actually never been contacted, so maybe it's too challenging ;-)
[0] Invalid MXes will be rejected by my hoster, but there's nothing like a Bayesian spam filter.
Given your profile page contents, ChatGPT correctly deducted your email. Luckily it is too resource-intensive for spamers to run an LLM on each web page scraped.
Interesting. We are in the process of judging how much to raise price for https://serpapi.com plans. Costs of everything (including servers, IPs, and other infra costs) have indeed increased at equal performance. The ratios I am toying with were more like around 3x max (and we thought it was too aggressive already) not 5x though.
As a customer, please give enough notice and some initial feedback, if you 3x our costing(or even 2x), we’ll be forced to find an alternative or shut down the feature that relies on your API.
I remember when Google did this for Maps. What a glorious day that was in the OpenStreetMap community. Big increase in "how do I host a tile server / geocoder" questions coming in.
Where would Bing API users move to, I wonder. How does it compare to Google pricing? Or ahrefs and petal, which are hitting my websites day and night? (Edit: tail -f access.log right now shows that ahrefs and petal are indeed going at it, no idea what products they offer though.) Yandex recently also upped its crawling game, has me wondering if they are making a copy of the web a la Internet Archive in case of more international, eh, operations.
Oh, I didn't know that they stopped. There used to be a pretty expensive API for search queries which I kinda wanted to use to make an improved interface (dark theme, keyboard shortcuts) in like 2012.
Oh well. This is what happens when you depend on someone else's API. There is nothing game changing about an AI SaaS product with it hallucinating its results that will justify the price hike.
This will just about ruin this current AI hype, like how OpenAI started charging the GPT-3 API.
The work of a master craftsman using the most basic hand tools is still leagues ahead of any amateur with advanced tools, automation, fancy CNC, and lasers. Mass production is where the tech would favour that amateur. We are surrounded by cheap mass-produced disposable products and are well aware of their low quality. No gourmet chef should be threatened by the efficiency of McDonald's. Fortunately, our software systems are custom artifacts with zero marginal cost of reproduction.
AI will make competent programmers more productive and take expert programmers to new levels.
One transformation will come when a single developer can do the work of a team of ten today. Like solo content creators who no longer need large studios to finance and organise massive production teams and expensive equipment, we'll see more independent developers and small teams creating innovative software with some artistic character. Without the drag of management and upfront financing, the threshold for success will be much lower.
Another will be a market for lemons. Employers will have to find ways to establish competence at building high quality, understandable and maintainable systems. AI can already pass Leetcode-style tests so these are clearly not the right measures.
This is my best optimistic take on things. The rest of the time I'm worried the AI is coming for our jobs.
116 comments
[ 0.17 ms ] story [ 163 ms ] threadSeems accurate.
Wonder where’s the customer with a preference for a 4x-10x price hike.
Also interesting how the LLM API pricing jumps from $28 to $200 per thousand transactions when you cross the 1M requests per day threshold.
Or the $28 is their lossleader and the $200 closer to their desired margin.
It's only natural that when your search product has been making headlines for being untrustworthy, inaccurate, and creepy/threatening that you'd start charging excessive amounts for access to it.
I'm sure all their data on "customer consumption preferences" indicate that consumers prefer paying higher and higher prices to consume products of questionable value.
Or it's an attempt to kill off DuckDuckGo and drive people towards using Bing directly.
It's definitely expensive to run an LLM behind every single search query, but it doesn't seem like these changes are related to that, with the exception of the "Information on a new way to use the Bing APIs: Bing API Pricing with Large Language Models " section below the rest of the pricing changes.
I didn't even consider that. That would be truly evil of them.
That seems odd. DDG is a for-profit company. Microsoft is a for-profit company. Why is DDG entitled to service?
(Why would it not be profitable any more? Maybe it never was, and they've been selling the API access at below cost to try to effectively buy the query stream and get scale that way. Maybe they're planning to massively improve search quality, and the costs will go up.)
Bing has, what, 3% market share? Why in the world would they invest energy in trying to help someone else carve out 1%?
Reminds me of when color laser printers first came out and on the cheap ones instead of pages per minute it was minutes per page.
> Our searching includes anonymized requests to traditional search indexes like Google and Bing as well as vertical sources like Wikipedia and DeepL or other APIs. We also have our own non-commercial index (Teclis), news index (TinyGem), and an AI for instant answers.
Of course using something good for free is not sustainable. So we may need to pay for search...
They also have a few nice extra features like being able to customize your own index by marking certain domains up-ranked or down-ranked (or removed entirely), depending on what you like, and you get full control over that.
There are only a few companies that I advocate so highly for, but I feel like they’re just so good…
(Edit: maybe they don’t rely as heavily on their own index actually? Now I’m not so sure. Will look into it to fact check myself)
Perhaps they have an agreement with Bing, perhaps they scrap Google search using 4G modems, perhaps they do an amazing job using Common Crawl.
They use both Bing and Google indexes. It is why the cost per search is relatively high.
This change will impact them and it's currently a discussion topic on their Discord as well.
Pretty sad. I just started using it a week ago and am still planning on not going back.
https://dkb.blog/p/brave-search-interview
Finally no more Pinterest! I'm still debating Kagi, but chances are high I will subscribe.
I think that DDG has its own crawler and index that it uses to "supplement" the results. I wonder if that is fully fleshed out or just something tiny and inconsequential.
They also stopped letting me filter results I don't want. "office -microsoft" for example returns pages of microsoft office links. "apple -id" returns results with "apple id" in the title and body.
I don't know if this is due to DDG specific changes, or if bing has gotten that much worse, but either way I'm having to G! a whole lot more than I'd care to.
> You do not need to take any action; this change will take effect automatically and be reflected in the next billing cycle after the change is activated.
This is exactly why editorialising is bad. I don't want people to start inserting superlatives into their submissions to get more attention.
You could editorialise and say "unjustfied" or "long overdue" or "uninteresting" but the FACT of the thing is that it is a massive price rise with zero editorialising about how /your/ or /I/ or anyone feels about that.
What do you see as editorialising there?
> Please use the original title, unless it is misleading or linkbait; don't editorialize.
"Massive" is an assessment inserted by the submitter. I agree with the assessment, but I'd rather see that in a comment than the title. Alternatively, use the word "massive" in the title of a blog post, and submit that.
The original title IS misleading. The price has gone up a lot, water is wet. Massive describes to the reader accurately what this is without saying anything about it being good or bad, clever or stupid, a sign of something else or anything at all non factual.
The original title is misleading. Any title that does not include price increase that is large is also misleading.
I think you are all simply imposing your feelings about the price hike on the factual title and then saying ooh, editorial, but that isn't there, it's in your own mind.
But hey, maybe I'm wrong, maybe you can sub edit a title that describes the change in price somehow more accurately. Let's see that.
No points for describing an increase as an update, that's really doublespeak and very poor unless you're Microsoft PR or something.
That said, conventions help make this transparent, like square brackets for changed or inserted words, and ellipses for deleted words
That is massive to me. If I were a Bing user, I would definitely look at dropping this. If MSFT is this confident about LLMs then maybe it will be worth it, but there's lots of work to do, and it needs to be done pretty quickly
But overall I'd like to see it. Maybe some pre-index struture? Just a large db of a page metadata? So that search engines can add their ranking on top of it.
I wonder if "not an index but a large database with raw HTML of pages" would cost less if you could get content providers to maintain a distributed cached copy of their content at their own expense?
Let's see how ecosia and DDG take this price increase.
I've long suspected Cloudflare itself commissions DDOS attacks just to sell their own services and that they attack rivals too. Look at one of the live maps of cyber-attacks some time, you'll see a lot originate and are destined inside the US, and that they mainly come from cloud providers. Then just imagine how cute you'd feel about yourself if you figured out how to use the cloud to burn everybody who didn't pay you protection money or enter into your free tier where you merely own the TLS certs. I think they got a bit too cute just like we've seen before with the dragon infitity logo signature the NSA uses.
[0] Invalid MXes will be rejected by my hoster, but there's nothing like a Bayesian spam filter.
Where would Bing API users move to, I wonder. How does it compare to Google pricing? Or ahrefs and petal, which are hitting my websites day and night? (Edit: tail -f access.log right now shows that ahrefs and petal are indeed going at it, no idea what products they offer though.) Yandex recently also upped its crawling game, has me wondering if they are making a copy of the web a la Internet Archive in case of more international, eh, operations.
This will just about ruin this current AI hype, like how OpenAI started charging the GPT-3 API.
One transformation will come when a single developer can do the work of a team of ten today. Like solo content creators who no longer need large studios to finance and organise massive production teams and expensive equipment, we'll see more independent developers and small teams creating innovative software with some artistic character. Without the drag of management and upfront financing, the threshold for success will be much lower.
Another will be a market for lemons. Employers will have to find ways to establish competence at building high quality, understandable and maintainable systems. AI can already pass Leetcode-style tests so these are clearly not the right measures.
This is my best optimistic take on things. The rest of the time I'm worried the AI is coming for our jobs.
You had me in the first half, not gonna lie.