Best YouTube API Providers in 2026 (Quota, Credits and Real Cost)

Six companies will sell you YouTube data, and no two of them charge for the same thing. One bills per request, one per record, one per video, one per twenty results, two per credit — and Google, which owns the data, does not sell it at all: there is no rate card and no checkout anywhere on the YouTube Data API. Put those headline numbers in one column and the column is meaningless. This post prints every published rate beside the unit it is actually charged in, shows the arithmetic for every figure that is our own division rather than something a vendor prints, and says plainly where somebody beats us. Checked 2026-09-16.

I read every rate below off the vendor's own live pricing page on 16 September 2026 and linked each one inline. I ran no benchmark and I make no claim about anyone's speed or success rate — this is a pricing, billing-unit and coverage comparison built from published documentation. Serpent is the API this site sells, and it appears here as one row among several.

How I compared these YouTube API providers

Four rules, and they are the reason this table looks different from the others you will find on this query.

• Entry rate and lowest published rate are separate columns. Quoting one vendor's volume floor against another vendor's entry price is the single most common dishonesty in this category. Both are shown, and so is what the floor costs every month before you have made a single call.
• A rate the vendor does not print is marked as our arithmetic. Two rows in the table carry rates that are a plan price divided by a plan volume rather than a number anybody publishes. Both are labelled, and every division is written out underneath so you can check it.
• If a vendor never says how many items one unit returns, the cell stays empty. I would rather leave a gap than fill it with an assumption that flatters somebody. That rule costs Bright Data a per-search figure on this page, and it should.
• Exclusions are findings, not omissions. Everything I left out is named in the caption below the table and again in its own section, with the reason.

The job I had in mind while reading is the ordinary one: run a keyword search on YouTube, then pull metadata and statistics for the videos it returns, on a schedule. If your job is subtitles, comment threads or channel archaeology, the coverage table further down matters more to you than the price table does.

The published-price table

Every rate here is the vendor's own published figure for the vendor's own billing unit. The units are not the same, which is exactly why the second column exists and why the raw numbers cannot be ranked against each other without it.

Prices checked 2026-09-16, each read off the vendor's own live pricing page on that date; every provider name in the first column links to the page the figure came from. Two rows above carry our own division rather than a number the vendor prints, so here is the working. Scrapingdog's LITE plan is $40 a month for 200,000 request credits and a YouTube Scraper API request spends 5 of them, so $40 buys 40,000 requests, which is $1.00 per 1,000 requests; Scrapingdog's largest published plan is $30,000 a month for 1,100,000,000 credits, or 220,000,000 YouTube requests, which is $0.14 per 1,000 requests. SerpApi's entry plan is Starter, $25 a month for 1,000 searches, so $25.00 per 1,000 searches; the SerpApi plan behind $1.96 per 1,000 searches is Cloud 54M, at $106,050 a month for 54,000,000 searches. Serpent's own rates are its published rate card, also on the pricing page. Re-verify before you commit a budget — rates in this category change without notice.

Two things survive that table. The first is that the unit matters more than the number: a vendor charging $1.50 per record and a vendor charging $0.60 per twenty results are not within a factor of two of each other on the same job, whatever the headline suggests. The second is that the cheapest figure on the page is not for sale — it belongs to Google, and Google's YouTube quota is allocated rather than priced. If you want the same exercise done across web-search vendors rather than YouTube ones, it is in the SERP API pricing comparison, and the costs that do not appear on anyone's rate card are in the hidden-costs breakdown.

Google's own API, and what its quota now allows

Start here, because it is the row most comparisons get wrong — including, until today, two posts on this site. The YouTube Data API v3 getting-started page states the allocation in one sentence: "Projects that enable the YouTube Data API have a default quota allocation of 100 search.list calls, 100 videos.insert calls, and 10,000 units per day combined for all other endpoints." The same page prices a search: "A search query costs 1 unit."

That is a different structure from the one the internet has been quoting for years. The old arithmetic was a flat 10,000 units a day with search.list priced at 100 units, which divided out to 100 searches. The daily search number is unchanged, but the route to it is not, and the practical consequence is real: your searches no longer eat the pool that your video, channel and playlist lookups draw from. Under the old sum, a hundred searches consumed the entire daily allocation and left nothing for anything else. Under the published one, 100 search.list calls and 10,000 units for everything else are separate budgets. If you have been rationing metadata lookups to protect a search budget, you can stop.

What has not changed is the commercial position, and it is the reason this row sits last in the table rather than first. Google publishes no rate card, no per-1,000 figure and no checkout for YouTube data. The page's own remedy for running out is an application: "If you reach the quota limit, you can request additional quota by completing the Quota extension request form for YouTube API Services." A form is not a price. Writing "$0" or "free" in a comparison table for a product that cannot be bought at any published price is the sort of tidy cell that sends somebody into a procurement conversation they did not budget for, so the honest entry is no published price, additional quota by application only.

None of this makes the official API a bad choice. If your volume fits inside the free allocation, it is unbeatable on price, it is the only option with write access, and it is the only one that can read private channel data through OAuth. It costs you a Google Cloud project, OAuth credential management and a token refresh path instead. Our complete guide to the YouTube API walks the same lookups both ways if you want to see the code side by side.

What one billing unit actually buys

The headline number is the easy half of a YouTube API price. What it is charged per is the half that decides the invoice, and there are five different answers on this page.

Per call — up to 50 rows for one charge

Serpent bills one unit per API call regardless of how much that call returns. A YouTube search takes a num of 1 to 50, and the batch video endpoint takes up to 50 ids at once; a call returning fifty rows costs exactly what a call returning one costs. Serpent's rate is $0.20 per 1,000 calls on the entry band, $0.02 on Growth and $0.01 on Scale, and those bands unlock on a one-time deposit rather than a monthly subscription.

Per 20 results — the unit that is not a request

DataForSEO's YouTube pricing page is explicit that on the Search and Comments engine types "you will be billed for every 20 results". A 100-result YouTube search is therefore five billing units, not one: at DataForSEO's Standard queue rate of $0.60 per 1,000 units that search costs $0.003, and on DataForSEO's Live mode at $2.00 per 1,000 units it costs $0.01. That is still cheap in absolute terms, but it means the rate you compare against a per-request vendor has to be multiplied by however many twenties of results you actually pull.

Per record — and nobody says how many records a search makes

Bright Data bills its YouTube Scraper API per record, at $1.50 per 1,000 on pay as you go, dropping to $1.30 per 1,000 for records beyond the 384,000 included in its $499-a-month Scale plan. A record is one row of output. Bright Data's YouTube product page lists what its endpoints cover — videos and posts by URL, keyword search, channel, hashtag, explore and podcast, plus channels and comments — but it does not say how many records one keyword search produces. There is no honest way to turn a per-record rate into a per-search rate without that number, so I have left the cell empty rather than pick a flattering value for it.

Per video — and the page disagrees with itself

The Apify actor streamers/youtube-scraper bills per video scraped, and its page headline reads from $2.40 per 1,000 videos, while the Apify FAQ further down that same page states that it currently costs $5.00 per 1,000 videos. Those two figures sit on one vendor page and they contradict each other; no per-plan table was readable on 2026-09-16, so which plan unlocks which figure is unknown. I am printing both rather than picking the one that makes the comparison tidier. Note also that Apify's actor discounts are tied to platform plans, and Apify publishes no Platinum or Diamond plan price at all, so any rate below the printed ones is a sales conversation and not a published price.

Per credit — where the surprise usually lives

Scrapingdog sells credits rather than requests, and the conversion is the whole story. Scrapingdog's LITE plan is $40 a month for 200,000 credits, which works out at $0.0002 a credit, and a YouTube Scraper API request spends five of them — so the plan buys 40,000 YouTube requests, not 200,000. Scrapingdog's YouTube Transcript API spends one credit instead of five, so the same balance goes five times further on transcripts than on the scraper; they are different endpoints with different economics and should never be quoted as one rate.

HasData does that arithmetic for you, which is a genuine usability win: its pricing page prints a dollar figure per 1,000 requests for every API it sells, and HasData's YouTube Scraper API sits in the same column as its Google SERP API at $2.46 per 1,000 requests on Startup, $0.99 on Basic and $0.69 on Growth. You never have to know the credit cost to budget it.

The multipliers that move the bill

Every vendor on this page has at least one number that is not on the rate card and changes what you actually pay. These are the ones worth knowing before you sign anything.

DataForSEO: two of its four YouTube engine types cost three times the table price

DataForSEO's YouTube SERP API is really four products — Search, Comments, Video Info and Subtitles — and it does not price them alike. The pricing page tells you to multiply the prices in its table by 3 for Video Info and Subtitles. On DataForSEO's Standard queue that is three times $0.60 per 1,000 units. So the rate you budgeted off the headline covers keyword search and comment threads only; pulling one video's metadata or its subtitle track costs three times as much per unit. If your workload is mostly video lookups rather than keyword searches — and for most competitor-tracking jobs it is — price it at the tripled rate from the start. This is not a footnote; on a metadata-heavy pipeline it is the difference between two budgets.

DataForSEO: a minimum you pay before you use anything

DataForSEO's minimum payment is $50. That is not a price per call and it is not a subscription, but it is money that leaves your account before the first request, and on a small evaluation it dominates the actual usage cost.

Bright Data: included records and additional records are two different rates

Bright Data's Scale plan is $499 a month with 384,000 records included, and the $1.30 figure is the rate for records beyond those. Spread across the included allowance the plan works out slightly under that, and if you use materially less than 384,000 records a month your effective rate is materially worse than either published number. The plan price is the floor you pay whether you call it or not.

Apify: the actor start, and the tier you cannot see

Apify's actors are paid per event on top of a platform plan, and the store discount that lowers an actor's per-event rate is attached to the plan you are on. Since Apify publishes no Platinum or Diamond plan price, the bottom of any actor's published discount ladder is not a rate you can sign up for. Treat the printed figure as the price and anything below it as a quote.

Scrapingdog and HasData: credits are not requests

Covered above, and worth repeating because it is the most common budgeting error in this category: a plan advertised in credits buys a fraction of that many requests, and the fraction depends on which endpoint you call. Read the credit table before the price table.

What each option actually returns

Price is the part every comparison covers. What comes back is the part that decides whether the cheap option is usable, and it is where the real differences sit. This table lists only the surfaces each vendor names on its own product page — not what I assume it can do.

Coverage checked 2026-09-16 from each vendor's own product and documentation pages. This is a list of what each page names, not a benchmark: I did not call anybody's API and I make no claim about how complete or how fresh any of these responses are in practice. Serpent's four YouTube endpoints and their exact response shapes are published in the endpoint documentation.

One field-level habit is worth calling out because it is rarely in anyone's marketing. On Serpent every YouTube response carries the same keys in the same order with the same types on every call, and a field we cannot state exactly comes back as its documented empty value rather than as a missing key or an invented number — "" for a description we did not get, 0 for a view count we did not get. That distinction matters if you are writing these into a statistics column, because a 0 that means "not available" and a 0 that means "zero views" are very different rows in a report. When a call returns fewer videos than the ids you sent, the response carries a delivery block naming what was requested and what came back, so a short answer is visible rather than silent. If you are building a tracker on top of any of these, the shape of a partial answer is worth testing before you pick on price — the competitor channel tracking walkthrough shows what that looks like in a real daily run.

What I left out, and why

Silent pruning under a comparison heading is selection bias, so here is everything that could have been in the table and is not.

• Traject Data / SerpWow. Its documented engine list is google, bing, yahoo, baidu, yandex, naver, amazon and ebay. There is no YouTube engine, so there is nothing to price.
• Scrapingdog's YouTube Transcript API. It exists and it costs one credit a request rather than five, but it is a different endpoint returning a different thing, and quoting it in the same row as the scraper would make Scrapingdog look cheaper on a job it does not do. It gets a mention above and no row here.
• Other Apify YouTube actors. The Apify store carries several, written by different authors at different prices, and an actor's price is set by whoever publishes it. Only streamers/youtube-scraper is priced here, because it is the one whose published figures I read on the day. Do not assume any other actor charges the same.
• Contact-sales tiers everywhere. Bright Data, SerpApi, HasData and Apify all have a custom tier with no published price. A rate somebody could negotiate cannot be bounded from outside, so none of them appears as anyone's floor. Every "lowest" in the table is the lowest published rate.
• Any vendor whose YouTube pricing I could not read. If a page did not state a figure on 2026-09-16, the vendor is absent rather than estimated. An unsourced number in a comparison table is worse than a missing row, because nobody can check it.
• Google's videos.insert allocation. It is in the quota sentence and in the free-allowance column for completeness, but it is an upload budget. It has nothing to do with reading data and should not be counted as read capacity.

Which option fits which job

Nobody's paid rate on this table is lower than ours at either end of the ladder, and I would rather say where that stops mattering than pretend it settles everything.

• Low volume, non-commercial, and you can live with OAuth? Google's own API, and it is not close. It costs nothing, 100 searches a day is plenty for a side project, and it is the only option that can write. The price of admission is a Cloud project and a credential refresh path.
• You need subtitles or transcripts? We do not return them, and that is a straight loss for us on this job. DataForSEO sells Subtitles as one of its four engine types, SerpApi sells a YouTube Video Transcript API, and Scrapingdog sells a transcript endpoint at a fifth of the credit cost of its scraper. Pick on coverage, not on our rate card.
• You need comment threads at volume? Again not us. DataForSEO prices Comments on the same table as Search, Bright Data lists comments as a scraper endpoint, and Scrapingdog sells a YouTube Comment API. Any of those beats an option that does not have the surface at all.
• Large one-off harvests of videos from a handful of URLs? The Apify actor's stated ceiling of up to 20,000 videos per URL is built for that shape, if you can live with the price ambiguity on its own page.
• Steady, high-frequency read traffic on search, video, channel and playlist? This is what our rate card is built for: one key, no Cloud project, no daily unit ceiling to ration, up to 50 rows per charged call, and a band that drops to $0.01 per 1,000 calls on a one-time deposit rather than a monthly plan. If you want the wider platform rather than YouTube alone, the same key covers the rest of the social media APIs and the video search API across engines.
• Cross-platform video monitoring rather than YouTube alone? YouTube data on its own will not tell you where a video ranks in search. That is a different endpoint and a different question, covered in video search competitor monitoring.

FAQ

How much does Google's YouTube Data API v3 cost?

It has no price. Google publishes no rate card, no per-1,000 figure and no checkout for the YouTube Data API v3. Its getting-started page states a default quota allocation of 100 search.list calls, 100 videos.insert calls, and 10,000 units per day combined for all other endpoints, and says that if you reach the quota limit you can request additional quota by completing the quota extension request form. The honest way to write that in a pricing table is no published price, not zero, because the thing you would be buying is not on sale. Checked 2026-09-16.

How many YouTube searches a day does the free quota allow?

One hundred, and the arithmetic behind that number has changed. Google's getting-started page now gives search.list its own allocation of 100 calls a day and prices a search query at 1 unit, while the 10,000 units a day are for all other endpoints combined. The older and widely quoted version of the sum, 10,000 units divided by 100 units per search, reached the same 100 searches by a route that no longer matches what Google publishes. The practical difference is that your search budget and your everything-else budget are now separate pots, so 10,000 units of video, channel and playlist lookups no longer compete with your searches. Checked 2026-09-16.

Which YouTube API provider publishes the lowest rate?

Scrapingdog, at $0.14 per 1,000 requests, and reaching it costs $30,000 a month. That figure is our own division of Scrapingdog's largest published plan, $30,000 a month for 1,100,000,000 credits, at the 5 credits a YouTube Scraper API request spends. Among rates you can reach with no monthly commitment at all, DataForSEO's Standard queue at $0.60 per 1,000 billing units is the lowest third-party figure on the table, and Serpent's YouTube endpoints are $0.20 per 1,000 calls on the entry band. Checked 2026-09-16.

Why do two YouTube API prices that look alike cost very different amounts?

Because the word per is doing different work in each one. On Serpent a unit is one API call that returns up to 50 rows. On DataForSEO's YouTube SERP API a unit is 20 results, so a 100-result search is five units. On Bright Data a unit is one record. On the Apify actor streamers/youtube-scraper a unit is one video. Two vendors quoting the same number per 1,000 can therefore differ by a factor of fifty on the same job, and that is before anybody's multipliers. Checked 2026-09-16.

Does any YouTube API charge extra for video details?

DataForSEO does, and it is the largest multiplier on this page. Its YouTube pricing page prices four search engine types and tells you to multiply the prices in its table by 3 for Video Info and Subtitles, so a video-metadata job costs three times what the headline rate suggests. Serpent charges one price for all four of its YouTube endpoints, and the batch video endpoint accepts up to 50 ids in a single charged call. Checked 2026-09-16.

Is there a free YouTube API?

Several of the options here carry a standing free allowance, and the largest of them is Google's own. The YouTube Data API v3 gives every project 100 search.list calls a day and 10,000 units a day combined for all other endpoints at no charge, which makes it genuinely the cheapest option on this page if your volume fits inside it and you are willing to run a Google Cloud project and OAuth. Bright Data publishes 5,000 free records a month on its YouTube Scraper API, HasData publishes 1,000 credits a month on its free plan, and SerpApi publishes 250 searches a month. Serpent gives every new account a free allowance that works on any endpoint, YouTube included. Checked 2026-09-16.

Related reading, if you are building rather than buying: the complete YouTube API guide covers the endpoints and the code, and competitor channel tracking turns them into a daily run you can schedule.