The Most-Liked Tweets and Most-Followed Accounts (And Why Every List You Find Is Stale)
The Most-Liked Tweets and Most-Followed Accounts (And Why Every List You Find Is Stale)
There is a structural problem with every "most-liked tweet of all time" article, including the ones ranking above this one: they were accurate on the day they were published.
A ranking is a photograph of a moving thing. The post that held the record in 2022 does not hold it now, and the list you are reading was probably last touched a year ago.
★So this page does two things: gives you the records as they stand, and shows you how to check them yourself — because the second one keeps working after this article goes stale too.★
The records, as of writing
Most-liked post: Chadwick Boseman's memorial announcement (August 2020) held the record for years at roughly 6.9 million likes, and it remains the benchmark later contenders are measured against.
Most-followed account: Elon Musk, currently north of 200 million followers — a position taken over from Barack Obama, who held it for most of the platform's history.
Most-reposted post: Yusaku Maezawa's 2019 giveaway, at over 4 million reposts, set by an explicit "repost to enter" mechanic.
⚠️ Every number above is approximate and drifting. That is not hedging — it is the actual state of the data, and any page presenting these as fixed figures is misleading you about how the platform works.
The useful skill is not memorising the current holder; it is being able to check — more on that below.
Why the rankings drift
Three forces move these numbers, and they work at different speeds:
1. Records get broken quietly. No announcement fires when a post passes a threshold. The record changes; the articles do not.
2. Old records inflate. ★A post from 2020 keeps accruing likes for years.★ Rankings that compare a 2020 post against a 2026 post are not comparing like with like — the older post has had six extra years to accumulate.
3. Accounts get deleted, suspended, or renamed. When a record-holding post disappears, ★most published lists never notice★ and keep citing something that no longer exists.
The compounding effect: a list assembled from other lists inherits every one of these errors. Much of what ranks for these terms is a copy of a copy.
Checking a record yourself
A record claim is just a number on a specific post, and that number is public. Verifying it is one call.
import requests
BASE = "https://api.socialapi.tech"
KEY = "your_api_key"
HDRS = {"X-API-Key": KEY}
def verify(tweet_id):
"""Current engagement on a specific post — the actual record check."""
r = requests.get(f"{BASE}/v1/tweet/info",
params={"id": tweet_id}, headers=HDRS, timeout=60)
if r.status_code == 404:
return None # ★deleted — a record nobody can hold★
r.raise_for_status()
return r.json()["data"]
t = verify("1234567890123456789")
if t is None:
print("post no longer exists")
else:
print(f"{t.get('like_count', 0):>12,} likes")
print(f"{t.get('retweet_count', 0):>12,} reposts")
print(f"{t.get('quote_count', 0):>12,} quotes")
★Note the 404 branch.★ Record-holding posts do get deleted, and when they do, ★every list still citing them becomes wrong at that moment★ — which is exactly the failure this check catches.
Comparing follower counts across accounts
The most-followed question is easier, because follower counts sit on the profile. One batched call covers a whole candidate list:
def ranking(usernames):
"""Follower counts for many accounts, one call, ranked."""
r = requests.get(f"{BASE}/v1/batch/user_info",
params={"usernames": ",".join(usernames)},
headers=HDRS, timeout=60)
r.raise_for_status()
people = r.json()["data"]
return sorted(people, key=lambda u: -u.get("followers_count", 0))
for i, u in enumerate(ranking(["elonmusk", "barackobama", "justinbieber",
"rihanna", "cristiano"]), 1):
print(f"{i}. @{u['username']:<16} {u.get('followers_count', 0):>13,}")
★This is the version that stays true.★ (One call for the batch rather than one per account.) Run it today and you get today's answer; run it next year and you get next year's — which is the difference between a ranking and a method.
The measurement problem nobody mentions
⚠️ ★"Most-liked of all time" is not a question the platform can answer.★
There is no global leaderboard, no endpoint that sorts every post ever published by likes, and no way to enumerate the corpus. Every "all-time" list is really a list of posts somebody already suspected were contenders, checked and sorted.
That has a practical consequence: a genuinely popular post nobody thought to nominate ★simply never appears on any list★. The rankings are not measurements of the platform — they are measurements of what people remembered to check.
So the honest framing is: these are the highest-engagement posts among those widely tracked. That is a much narrower claim, and it is the only one the data supports.
Questions people ask
What is the most-liked tweet of all time? Chadwick Boseman's memorial post, at roughly 6.9 million likes, is the long-standing benchmark. ★Verify before citing — records move.★
What is the most-liked tweet ever? Same answer, and the same caveat: no list is current by default.
Who has the most followers on Twitter? Elon Musk, currently above 200 million. Previously Barack Obama.
Who is the most-followed person on Twitter? Same as above. Individual accounts, not brands, dominate the top of this list.
What are the most-followed Twitter accounts? A small set of global celebrities and political figures, plus a few platform accounts. The order changes; the membership rarely does.
What is the most-retweeted tweet? Yusaku Maezawa's 2019 giveaway, over 4 million reposts — driven by a "repost to enter" mechanic.
Is the most-liked tweet still the same? Check rather than assume. That is the entire point of this page.
How do I check a record myself? One call for a post's current counts, or a batched call for follower rankings.
Why do different sites give different answers? Different snapshot dates. Most rankings are copies of older rankings.
Does Twitter publish an official leaderboard? No. ★There is no global ranking endpoint, which is why every list is hand-assembled.★
Can I find the most-liked tweet on a topic? Not exhaustively. You can search and sort what you retrieve, but that is not the same as ranking everything.
Can I get the most-liked tweet from an account? Yes — pull their posts and sort by likes locally. That is a bounded question, unlike "all time".
Why can the API not sort by likes globally? Sorting requires a corpus to sort. No one, including the platform's own search, exposes the full archive that way.
Do old tweets keep gaining likes? Yes, indefinitely. ★This is why age-mismatched comparisons overstate older posts.★
What happens to the record if the post is deleted? It stops being verifiable. Lists citing it become wrong without changing.
Which account gained followers fastest? Growth-rate records are even less reliable than totals — nobody samples frequently enough to establish them.
How often do these records change? Follower rankings shift monthly at the top. Like records break rarely, then hold for years.
Are follower counts accurate? They are accurate as counts. ★They do not tell you how many are active or real★ — a separate question.
What was the first tweet? Jack Dorsey's "just setting up my twttr", March 2006.
Can I see the top tweets from a specific day? Not as a ranking. You can search a date range and sort what comes back — how date search works.
Is there a most-viewed tweet record? View counts only became public in late 2022, ★so there is no comparable history before that★.
Do bots inflate these numbers? Some engagement is automated. It affects giveaway-driven records most, which is worth remembering for the repost record specifically.
Can I track a record over time? Yes — poll the post's counts on a schedule and store them. The platform keeps no history for you.
What is the most-followed brand account? Brands sit well below individuals. The gap is roughly an order of magnitude.
How many followers does the top account have? Above 200 million and rising. Any exact figure printed here would be wrong within weeks.
Can I compare two accounts directly? Yes — one batched call returns both, and comparison is local arithmetic.
Why is my number different from what a site shows? Caching. Many sites refresh rankings rarely, and some never refresh them at all.
Are these records region-specific? The global lists skew heavily toward English-language accounts. ★Regional records exist but are far less tracked.★
What about the most-quoted tweet? Rarely tracked, because quotes were counted separately only later. The data is thinner than for likes.
Can I build my own leaderboard? Yes, for a defined candidate set. Define the set, pull the counts, sort. That is an honest leaderboard.
Does verification affect follower counts? No. It changes visibility mechanics, not the count itself.
Do suspended accounts stay in rankings? They should not, but published lists routinely keep them.
What is the best way to cite a record? ★With the date you checked it.★ A record without a timestamp is an assertion, not a fact.
The short version
The records above are real and current as of writing, and that qualifier is doing genuine work.
★"Most-liked of all time" is not something the platform can answer — there is no global leaderboard and no way to enumerate every post.★ Every ranking is a hand-picked candidate list, which means a popular post nobody nominated never appears on one.
What survives that problem is a method rather than a list. Our API returns current engagement on any post and batched follower counts across many accounts, so verifying a record, or building a leaderboard over a set you define, is a short script. Read-only, flat price per call.
What no API offers: a sorted index of every post ever published. ★That endpoint does not exist anywhere, which is the real reason these lists disagree with each other.★
Related reading: what the engagement metrics mean · searching a date range · how many followers are real · pulling an account's full history.