The editorial argues Kagi is reframing a growth move as a philosophical pivot. The old paywall-first model required strangers to commit a credit card before evaluating the product; the free tier lets prospects taste it first, positioning Kagi as the default answer to 'alternative to Google' rather than a niche subscription.
Search is a scale business where crawling the open web costs tens of millions annually before rendering a single result. A user base in the low hundreds of thousands — Kagi crossed 50k paying users in 2024 — cannot underwrite a serious independent index, making the free-tier expansion a necessary funnel rather than a pure values play.
By submitting the changelog with 603 points of community uplift, the HN audience signaled that Kagi's core pitch — no ads, user-controlled ranking via lenses and site boosts, AI features layered on real results rather than a chatbot — still resonates. The paywall drop is framed as expanding access to that model, not abandoning it.
Kagi — the small, paid-only search engine that spent the last three years being the darling of HN power users — pushed a changelog entry titled Kagi Search for Everyone. The post climbed to 603 points on Hacker News, which, for a product changelog, is the kind of surface area most launches pay six figures to simulate.
The change itself is simple on the tin: Kagi is dropping the hard paywall that gated every query behind a $5–$25/month plan and opening the index to unauthenticated users with a limited free tier. The paid plans — Starter, Professional, Ultimate — remain, and the pitch for them hasn't changed: no ads, user-controlled ranking (`.lens`, site boosts, personalized blocklists), AI features bolted on top of real search results rather than a chatbot pretending to be one.
What's new is the posture: Kagi now wants to be the first result when someone types 'alternative to Google', not the last one after a trial of three. The site positions this as a mission statement — 'humanizing the web' — but strategically it's a growth unlock. The old model required a stranger to type their credit card in before they could evaluate the product. The new model lets them taste it first.
For six years Kagi's entire identity rested on one proposition: you are the customer, not the product. That phrase got a lot of mileage on HN, and it was more than marketing — it determined the ranking function. No ads means no incentive to privilege affiliate-stuffed SEO farms. No tracking means no behavioral profile shaping your results. The monthly bill was the moat.
The critique, mostly whispered, was always economic. Search is a scale business. Crawling the open web costs tens of millions of dollars a year before you render a single result, and a user base in the low hundreds of thousands — Kagi crossed 50k paying users in 2024 and has grown since — can't underwrite a serious index on its own. Kagi has been honest that it leans on a blend of its own crawler (Teclis, Tinygem) plus licensed results from Google, Brave, Marginalia, and Mojeek. That blend is fine at the scale of a boutique, but it's structurally dependent on competitors who could raise their API prices or cut access at any time.
Opening the door to free users changes the math in two directions. The optimistic read: more users means more signal for the ranking model, more data on which `.lens` filters people actually want, more leverage when negotiating index-licensing contracts. The cynical read: a free tier is a funnel, and funnels only work if a non-trivial fraction converts to paid — which forces the product to eventually make the free tier worse than the paid one in ways that may not survive contact with 'humanizing the web'.
Community reaction on HN tracked this split cleanly. The top comments were congratulatory — 'finally, I can send Kagi links to my non-nerd friends' — followed quickly by the predictable 'how long before this becomes DuckDuckGo', meaning: how long before ad pressure, VC pressure, or scale pressure forces a ranking compromise. Kagi's founder Vladimir Prelovac has been unusually transparent about finances in past AMAs (publishing revenue, costs, burn), which is why the community gives them more rope than they'd give a Silicon Valley startup making the same pivot. That trust is the real asset here, and it's not transferable to whoever runs Kagi five years from now.
The other context worth naming: the competitive landscape above and below Kagi has shifted. Perplexity and ChatGPT search have eaten the 'ask a question, get a synthesized answer' use case that used to anchor the top of Google's funnel. DuckDuckGo stopped being interesting around the time Bing results got worse. Marginalia is a hobby project. Brave Search is fine but entangled with a browser most devs don't use. Kagi is, for all practical purposes, the only paid search engine with a working product and a real user base — and it just decided that 'paid' was the wrong hill to die on.
If you already pay for Kagi, nothing changes today — your plan, your lenses, your blocklists, your API key all keep working. The interesting move is for the 90% of developers who've bookmarked `kagi.com` after reading a glowing HN comment and then never actually used it because committing to a monthly fee for *search* felt absurd. You can now test-drive it the way you test-drive any other tool, which is the barrier that was actually keeping Kagi off most devs' daily driver shortlist.
The practical workflow: set Kagi as your default engine in Firefox or Arc for a week, lean on the `!g` bang to bail out to Google when you need to, and pay attention to how often you actually need to bail. In my own usage the failure modes are predictable — hyper-local queries (restaurants, business hours), real-time news in the first hour of a story, and anything requiring Google's knowledge-graph scaffolding (sports scores, flight statuses). Everything else — docs, Stack Overflow, blog posts, GitHub code search, package manager pages — Kagi handles as well or better, mostly because it hasn't been optimized to death by SEO farms yet.
If you're building anything that depends on web search as a primitive — a RAG pipeline, an agent, a scraper — Kagi's Search API becomes newly relevant now that the consumer product is the funnel instead of the gate. The pricing on the API ($25/1k queries for the small plan) was always the kind of thing you'd evaluate against SerpAPI or Brave Search API, and the ranking quality has been consistently better in my own A/B tests on technical queries. The one caveat: API access remains paid, and there's no sign that changes — 'search for everyone' refers to the consumer web app, not the developer product.
The next twelve months will tell us whether Kagi can pull off the trick exactly one search engine has ever managed: scale a quality-first ranking function past the point where scale usually corrupts it. The optimistic case is that a free tier with real limits (query caps, feature gating, no API) stays philosophically consistent with the paid tier and funds a better index for everyone. The pessimistic case is that the gravitational pull of free users eventually forces the product toward ad surfaces or data licensing, and the HN comments five years from now read like today's DuckDuckGo threads. Either way, this is the first time since Google launched that a search engine with a genuinely different ranking philosophy has made a real bid for the mainstream — and it's worth setting it as your default long enough to find out which way it breaks.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.