Web search for AI,priced againstwhat it returns.
askFinz search runs on its own index and is used inside askFinz: the workspaces, the browser extension and the desktop app. Search APIs are what you would otherwise build against. This page puts both on one rate card, per 1,000 searches, and checks whether each returns the page text or only a link to it.
"web search for ai"/
├── a product you use/
│ └── askfinz: workspaces, extension, desktop
├── a search api, own index/
│ ├── keiro, brave, exa/
│ └── you.com, linkup, tavily, serpapi
└── a search api, relayed/
├── no index of its own
└── reaches only what rankedThree things worth knowing before you compare a rate card.
Two different products, one label
"Search API" covers services that hold their own index and services that relay a search engine's results back to you. Both are useful, and they are priced very differently for a reason — see the note under the table for which is which.
What the call returns changes the price you should compare
Some APIs charge extra to hand back the actual page text; others include it. Comparing headline rates without checking that is comparing different products — the table below matches on what you actually get back.
The cheap tier disappeared once already
Bing's Search API — the budget option a lot of this market was built on — was retired in August 2025. Its replacement is a different kind of product at a materially higher price, which is part of why the ground here keeps shifting.
Among index owners, this is the market.
| Provider | Per 1,000 searches | vs askFinz bundled |
|---|---|---|
askFinz — top-upextra searches, with the extension or a paid plan | $0.75 | — |
askFinz — with the extension or a paid plan1,000 a month included in what you already pay | $1.00 | — |
askFinz — standalonewithout the extension or a paid plan | $3.50 | — |
Parallelturbo and fast tiers. The rate covers the FIRST 10 RESULTS; further results are $1.00 per 1,000 on every tier, so a call returning 100 results works out near $91 per 1,000 requests. Basic and advanced $5.00 | $1.00 | about the same |
Keiropage text at their best tier; $4.00 and $7.20 below it. Ranked links alone $0.80, and a lighter tier at $0.25 | $2.40 | 2.4× more |
Keenablepay as you go; $1.00 only on dedicated capacity at 100 requests/second | $4.00 | 4.0× more |
Yepquery-relevant highlights at no extra cost; advanced tier $8.00 | $4.00 | 4.0× more |
Brave Search APIanswers tier $4.00 | $5.00 | 5.0× more |
Exa — answer | $5.00 | 5.0× more |
You.com Search API1–100 results per call | $5.00 | 5.0× more |
Linkupstandard depth; their deep tier is roughly ten times this | $5.00 | 5.0× more |
Tavilylargest published plan; $8.00 pay as you go, and advanced search costs double | $5.00 | 5.0× more |
Exa — standard searchfirst 10 results; +$1.00/1k per result after that | $7.00 | 7.0× more |
Amazon Bedrock Web Searchsemantic passages from Amazon's own index | $7.00 | 7.0× more |
Exa — deep searchdeep reasoning $15.00 | $12.00 | 12.0× more |
You will find cheaper rates than these. A separate group of services — SerpApi, DataForSEO, Serper, SearchAPI and others — quotes from about $0.60 per thousand, and they are worth understanding rather than dismissing: they don't hold an index of their own, they pass your query to a search engine and hand back what it returned. That is a genuinely useful thing to buy, and it is priced accordingly. It is also a different product from this one. They return the search engine's results — titles, links and snippets — not the text of the pages behind them, so they can only reach what the engine chose to show, and none of the material we hold because we read it from the source. Among services that run their own index, the rates above are the market.
What the model providers charge for web access. If you are giving an LLM the web rather than querying an index yourself, the meter runs inside whoever runs the model, and the published figures for the same verb run six times end to end: xAI’s web search tool $5.00 per 1,000 calls, Perplexity’s Search API $5.00 per 1,000 requests for raw results with filtering, Anthropic’s web search $10.00 per 1,000 searches, OpenAI’s $10.00 — $25.00 on non-reasoning models — Google’s grounding $14.00 per 1,000 grounded queries, and Mistral’s web search $30.00 per 1,000 calls. Every one of those is charged on top of the tokens the pages themselves cost you, with one exception: OpenAI’s $25.00 tier, where the search content tokens are free. Anthropic also adds 10% to Opus 4.7 and above token costs for US-only data residency, which base model pricing does not show. They are not in the table above because they are not an index you can query — they are the model’s access to one, and what comes back is results or a grounded answer rather than the passage and its source. Rates confirmed 7 September 2026.
The budget option is gone. Bing's Search API — what much of this market was built on — was retired in August 2025. Its replacement is a different kind of product at a considerably higher price, which is part of why search costs what it does elsewhere.
Rates read from each provider's own pricing page on 7 September 2026. Where a provider gets cheaper with volume, the comparison uses their best rate, not their entry price. Rates are matched on what the call returns: askFinz search returns the relevant passage from the page with every result, so where a provider charges separately for page text, that is the rate shown. Providers change pricing — check theirs before relying on this.
A few kilobytes a page is what makes the rate above possible.
Every search API in the table above that owns its index is carrying the same fixed cost: storage for everything it’s read. Keep that small and a low rate is sustainable; let it grow and it isn’t. This is the live figure from the running system.
What's the difference between askFinz and a search API like Tavily or Exa?
A search API is something you call from code you write. askFinz is a product you use: search, research agents, a browser extension and a desktop app on one sign-in, and it is not sold as an API. Both run on an index of the web, so this page compares what a thousand searches costs and what each one returns.
Why do some "search APIs" cost a fraction of the ones in this table?
Because they're a different product. A relay-style API holds no index of its own — it passes your query to a search engine and returns what came back. That's genuinely useful and priced accordingly, but it can only ever reach what the engine chose to show, and none of what an owned index reads directly from the source.
Is this list every search API on the market?
No — it's the ones that publish a clear per-query rate and run their own index, so the comparison is apples to apples. Vertex AI Search is a notable absence: its pricing is multi-dimensional enough that a single per-1k figure would misrepresent it either way.
Is there an askFinz search API?
No. askFinz search is used inside askFinz itself, in the workspaces, the browser extension and the desktop app, at the rates shown. There is no public API for the index, and the index is not sold or licensed as a separate feed.
Everyone else in that table sells you an endpoint.
A rate per thousand calls is the whole product for most of the rows above. Here it is one way in. The same index answers in askFinz Search, is read by the workspaces, and reaches the browser through the extension — so a price you can compare is attached to a product your team can actually open.
It answers, not just responds
askFinz Search reads this index directly. An answer arrives as the passage that addresses the question with the address it came from, rather than a list of links to go and read.
Try the searchThe workspaces read the same shelf
Research, Finance and the rest read the collections this index is filed into, so what you are comparing on price is also what your team opens on a Tuesday morning.
Every workspaceAnd it follows you into the browser
The extension puts the index and the models on the page you are already looking at — no tab to switch to, and on a compatible graphics card the models run on your own machine.
Get the extensionAnd the models are not metered either.
Multiple frontier models in one place, with no per-model subscription. The extension and the desktop app are one flat monthly price, and nothing you do inside them is metered — no per-token meter running while you work, and no cap to plan around. Put beside a search rate that bills per call and a model that bills per token, that is the part of the comparison the table has no column for.
Search, at the price of owning the index.
The table above is the rate a real account pays today, not a promotional figure. Try it directly, or read how the index underneath it is built and kept fresh.