Sessions that behave like people
Requests are paced per session, a rate-limited session rests on its own, and calls fail over to a healthy one. Each session keeps the browser identity it came from.
9 platforms · one schema →
Every public post.One social media API.
Profiles, posts, comments and search from X, YouTube, Instagram, TikTok, Facebook, Reddit, Telegram and the news. One key, one data shape, and you never pay for a call that fails.
Paste a link, we detect the platform. Real API responses, recorded 25 Oct 2026.
XYouTubeInstagramTikTokFacebookRedditBlueskyTelegramNews & web
One shape everywhere
The same fields
from every platform.
Every platform shows a post differently. Vanefeed returns the same fields every time: author, text, published_at, metrics, media. Your code is written once.
Real responses recorded 25 Oct 2026, shortened for display. A field a platform doesn't show comes back as null, never as an invented 0.
Under the hood
How we keep it running.
Tested every hour
A live test call goes to every platform every hour, with background checks every 15 minutes. When one breaks, its calls fail fast and free, and the status page says so before you have to ask.
Credits held, never overspent
Each call reserves its maximum, then charges only what came back. Failed and cached calls cost nothing, and parallel calls can never double-spend.
Newest first, deduplicated
Pages come back sorted by time, with repeats removed across pages, so a feed reads like a feed.
Why teams switch
Why the usual options
cost more than they say.
Platforms have priced up or shut their APIs. The third-party market that replaced them bills unpredictably, charges for failures and breaks without warning. Each number links to its source.
$5.00
per 1,000 post reads on the official X API, with no practical free tier.
Outstand, 2026 ↗~40%
of Instagram requests blocked on average across 11 scraping APIs tested (December 2025).
Proxyway via Phyllo ↗~100
YouTube searches a day on a default Data API project: 10,000 units, and each search costs 100.
Phyllo, 2026 ↗88
pricing complaints in G2's tagged negatives for the biggest scraping marketplace, its most cited problem.
Rewardio review ↗Failed calls are free
Errors, timeouts, rate limits and platform outages cost 0 credits. You pay only for successful responses.
Price before you call
Every endpoint lists its cost. Every response carries X-Credits-Cost, and dry_run=true quotes a call for free.
Credits never expire
Prepaid packs last forever. No monthly resets, no daily caps, no use-it-or-lose-it.
Breakage is public
Our status page shows live success rate and latency for every platform, from live test calls that run every hour.
Paste any link
A post, a channel, a subreddit or a news site: the playground and API work out the platform for you.
Scores show their work
Sentiment and intent arrive with the model name and the reason, so you never mistake an estimate for a fact.
Use cases
What you can build with it.
AI agents and RAG
Give your agent fresh posts, comments and transcripts as clean JSON, through MCP or plain HTTP.
/v1/youtube/searchBrand monitoring
Name your brands and competitors once. New mentions arrive by webhook with sentiment attached.
POST /v1/webhooksInfluencer research
Pull profiles, follower counts and recent posts across platforms in one shape to compare creators.
/v1/{platform}/profileMarket and trend research
Track which hashtags and topics are rising in the sources you care about, week over week.
/v1/trendsCommunity and support
Catch questions and complaints about your product on Reddit, YouTube and Telegram before they spread.
/v1/mentions/searchResearch and OSINT
Follow public Telegram channels, subreddits and news feeds and export what changed, with source links.
/v1/streamFast by design
First results in about two seconds.
The first page comes straight from the platform. While you read it, the next page is already loading, so scrolling never waits. Repeat requests come from cache and cost nothing.
Measured 3 Oct 2026 from a home connection; our servers sit closer to the platforms.
first pagenext page (prefetched)3–5 s target
What it costs
Estimate your monthly cost.
Lists cost 2 credits per 20 results. Packs are prepaid and never expire. Compare that with paying per read on an official API.
Estimates at published rates. Official X API at $5 per 1,000 post reads on pay-per-use.
Built for AI agents
Works with Claude, Cursor and other AI agents.
Every endpoint is a typed tool. Add the Vanefeed MCP server to Claude, Cursor or any MCP client, and your agent can search mentions, read profiles and follow threads on its own. Our docs are also published as llms.txt for AI assistants.
- Deterministic JSON in one schema for every platform
- Cost on every response, so agents can budget their own calls
- Free dry runs to plan before spending
{
"mcpServers": {
"vanefeed": {
"command": "node",
"args": ["/path/to/vanefeed-mcp.mjs"],
"env": {
"SW": "https://api.vanefeed.com",
"SW_KEY": "sk_live_your_key"
}
}
}
}A note from the team
We answer every message.
We built Vanefeed because getting public social data shouldn't mean surprise bills, broken scrapers and silence from support. So the rules are simple: you only pay for calls that work, your credits don't expire, and when a platform changes and something breaks, you'll see it on our status page and hear it from us.
Every access request and support email is read by one of us. We reply within one business day, usually much sooner. If something isn't right, tell us and we'll fix it.
M. Asif (Founder & CEO), Hardy Mate and Anastasia Smith (Co-founders) · hello@vanefeed.com