A pricing and feasibility scan of every IP-rotation route I could think of for a 35K-request, one-off-ish YouTube metadata job. The short version: don't buy proxies for this — use the official YouTube Data API v3, which finishes the work in 7% of one day's free quota. Everything else is here for completeness, and for the day the API isn't enough.
If you only read the cards: the official API wins on price, simplicity, and CAPTCHA risk. The proxy field has one cheap-enough option for this size of job (Webshare). Most of the rest is built for adversaries YouTube metadata isn't.
videos.list call. 35K IDs = 700 calls = 7% of one day's free quota. Returns the same fields you currently parse from yt-dlp.--proxy gateway URL. Decodo at $3.50/GB (~$6) is a reasonable second.--proxy gatewayHTTPS_PROXY and re-run.All costs scaled to ~35K requests ≈ 1.75 GB of response data. Prices verified 23 April 2026 from vendor pages or 2026-dated third-party reviews. Per-GB and per-request models are not comparable in marketing copy — the per-request column ($/req) absorbs that gap. Green rows are recommended for this job; honey rows are reasonable; muted rows are over-spec.
| Provider | Type | Headline price | Cost for 35K reqs | Free trial | yt-dlp setup | Notes |
|---|---|---|---|---|---|---|
| YouTube Data API v3 verified | Official API | Free, 10K units/day | $0 (700 units) | n/a | not via yt-dlp; requests + key |
Batches 50 IDs/call. Requires API key. |
| Webshare residential approx | Rotating residential | $1.40/GB (promo, ~half off) | ~$2.45 | 10 free proxies forever | --proxy gateway |
Promo flagged on site early 2026. |
| Decodo (Smartproxy) approx | Rotating residential | $3.50/GB PAYG, $2/GB at scale | ~$6 | 3-day, 100 MB | --proxy gateway |
115M+ IPs, good docs. |
| IPRoyal residential approx | Rotating residential | $7/GB entry, $1.75/GB bulk | ~$12 (entry) | none reliably | --proxy gateway |
Non-expiring traffic. |
| Oxylabs residential approx | Rotating residential | $4/GB PAYG, $3.49/GB Advanced | ~$7 | 7-day | --proxy gateway |
Aimed at enterprise volume. |
| SOAX residential approx | Rotating residential | $4/GB PAYG | ~$7 | $1.99 / 400 MB / 3 days | --proxy gateway |
Trial is paid but cheap. |
| Bright Data residential approx | Rotating residential | $5.88/GB starter, ~$8.40 at 10 GB | ~$15 | 7-day enterprise | --proxy gateway |
Largest pool; overkill here. |
| NetNut residential approx | Rotating residential | ~$15/GB entry, $1.59/GB enterprise | ~$26 (entry) | trial on request | --proxy gateway |
Worst small-volume pricing. |
| Webshare datacenter approx | Rotating DC | $2.99/mo for 100 IPs | $2.99/mo | 10 IPs free | --proxy rotation list |
DC IPs get more YouTube CAPTCHAs. |
| Bright Data Web Unlocker approx | Scraping API | $1.50–$12.60/CPM PAYG | $50–$440 | trial credit | swap URL, not yt-dlp | YouTube isn't a "hard" target; wasted spend. |
| ScraperAPI approx | Scraping API | ~$0.001/credit, JS = 5–25 credits | ~$50–$100 | 5K free credits | not yt-dlp; HTTP frontend | Premium target multipliers eat budget. |
| ScrapingBee approx | Scraping API | 5 credits/req default; $49/mo = 150K | ~$11.50 (no JS) | 1K free | not yt-dlp; HTTP frontend | Default JS render = 5× cost. |
| Zyte API unverified | Scraping API | $0.06–$16.08 / 1K successes | $2–$560 | $5 free | not yt-dlp; SDK | Tier opaque without account. |
| Apify YouTube actor approx | Pay-per-result | $0.0005/video (apidojo) | ~$17.50 | $5/mo free | actor run, not yt-dlp | Actors range $0.0005–$0.0024/video. |
This is the right answer for the job and it's worth belabouring, because most people reach for proxies first.
GET https://www.googleapis.com/youtube/v3/videos?part=contentDetails,snippet&id=ID1,ID2,…,ID50&key=…contentDetails.duration (ISO 8601) and snippet.channelTitle / snippet.channelId. Direct match.The cost-per-request gap between this ($0) and the cheapest paid proxy ($2.45 at Webshare) is small in absolute terms but infinite in ratio. The real benefit of the API is no rate-limit surprises and no CAPTCHA risk, ever.
All of these work with yt-dlp via yt-dlp --proxy http://user:pass@gateway:port — the provider rotates exit IPs server-side per request, so no proxy list is needed. yt-dlp accepts HTTP/HTTPS/SOCKS5 proxies natively (--proxy); no built-in --proxy-file exists yet.
These bundle proxy rotation + headless browsing + retries. You don't need any of this for YouTube metadata. They're priced for sites that fight back hard (Amazon, Google SERPs, Cloudflare-protected). Listed for completeness.
All four are worse versions of the YouTube Data API.
The DIY answer that actually works is the YouTube Data API. Everything else here is a worse version of it.
You have ~35K IDs, 16 workers, no current blocks, and a one-off-ish workload. Don't buy proxies. Two real options:
videos.list, parse contentDetails.duration (ISO 8601 → seconds) and snippet.channelTitle. Done in one HTTP-bound run with no rate-limit risk at all (700 of 10,000 units). The right call if this will be re-run periodically or if you want zero block risk.If you genuinely want a proxy as a fallback safety net for the current yt-dlp run: Webshare residential is the only one priced low enough to make sense at this scale (~$2.45 expected). Set it as HTTPS_PROXY only when retrying failed IDs; don't route all 35K through it.
What I'd flag: cost-per-request math punishes you for tiny responses. A 50KB metadata response on a $5/GB residential plan costs ~$0.00025 — but on a $1/CPM scraping API it's $0.001, 4× more. The two pricing models look comparable in marketing copy and aren't.
Where each number sits on the verified→guess axis. Vendor pricing pages change; promo rates (especially Webshare's $1.40/GB) can end without notice — re-check before purchase.
youtube.com — couldn't verify which of the 5 tiers it falls into without an account; the cost range $2–$560 reflects that uncertainty.Pricing pages and one yt-dlp issue. All accessed 23 April 2026.