Twitter search operators in 2026

#How Twitter search operators work

A search operator is a keyword with a colon, such as from: or lang:, or a symbol such as quotes or a minus sign, that turns a plain search on X into a filter. You type it into the search box on x.com or in the app, next to your ordinary words, and X returns only the posts that meet every condition.

To build a query by clicking instead of typing, open the free Twitter Advanced Search tool. Every filter you set there writes its operator into the query, and it shows the first 20 matching posts. X's own form at x.com/search-advanced covers words, accounts, places and dates, and it opens only for signed-in accounts (X Help Center (opens in a new tab)).

Seven rules decide whether a query works:

• A space means AND: vector database finds posts that contain both words, in any order.
• OR must be in capitals. A lowercase "or" is searched as a word.
• A minus sign in front excludes a word, a phrase or an operator: -crypto.
• Double quotes match an exact phrase and stop X from correcting the spelling.
• Parentheses group terms. AND binds tighter than OR, so cats OR black dogs reads as cats OR (black dogs).
• No space after the colon: from:nasa filters by account, while from: nasa breaks the operator.
• Operators ignore letter case, and X stops returning results after about 22 of them in one query.

#Every Twitter search operator, by what it filters

The tables below hold the operators that X's web search accepts, grouped by what they filter. X documents only the basics in its help pages; the rest come from years of public testing collected in Igor Brigadir's reference on GitHub (opens in a new tab), so try a rarely used operator on a small query before a report depends on it.

Words, phrases and logic

Words and phrases match the post's text, the author's name and handle, and the full form of shortened links.

• word word

  Finds

  Posts with both words

  Example

  nasa esa

• OR

  Finds

  Posts with either term

  Example

  rust OR golang

• "phrase"

  Finds

  The exact phrase, no spelling fixes

  Example

  "state of the art"

• -word

  Finds

  Posts without the word

  Example

  jaguar -car

• ( )

  Finds

  Grouped terms

  Example

  (rust OR go) compiler

• "a * b"

  Finds

  Any one word in the gap, inside quotes only

  Example

  "best * for startups"

• +word

  Finds

  The word exactly as typed

  Example

  +radiooooo

• #tag

  Finds

  A hashtag

  Example

  #buildinpublic

• $TICKER

  Finds

  A cashtag

  Example

  $NVDA

• ?

  Finds

  Posts that ask a question

  Example

  crm ?

Accounts

Account operators take the handle without the @, except the mention operator itself.

• from:

  Finds

  Posts written by an account

  Example

  from:nasa

• to:

  Finds

  Replies sent to an account

  Example

  to:nasa

• @handle

  Finds

  Posts that mention an account

  Example

  @nasa -from:nasa

• list:

  Finds

  Posts by members of a public List, by its ID

  Example

  list:1849212213712385091

• filter:verified

  Finds

  Posts from verified accounts

  Example

  robotics filter:verified

• filter:blue_verified

  Finds

  Posts from X Premium subscribers

  Example

  robotics filter:blue_verified

• filter:follows

  Finds

  Posts from accounts you follow; signed-in x.com only

  Example

  pricing filter:follows

• filter:social

  Finds

  X's pick of your wider network; Top tab, signed in

  Example

  rust filter:social

Engagement

Engagement operators set a floor on likes, reposts or replies, and a minus sign in front turns the floor into a ceiling. Counts above about 1,000 are approximate, so results close to a high threshold can be slightly off.

• min_faves:

  Finds

  At least N likes

  Example

  min_faves:100

• min_retweets:

  Finds

  At least N reposts

  Example

  min_retweets:50

• min_replies:

  Finds

  At least N replies

  Example

  min_replies:20

• -min_faves:

  Finds

  Fewer than N likes

  Example

  -min_faves:10

• -min_retweets:

  Finds

  Fewer than N reposts

  Example

  -min_retweets:5

• -min_replies:

  Finds

  Fewer than N replies

  Example

  -min_replies:3

• filter:has_engagement

  Finds

  Posts with any like, reply or repost

  Example

  -filter:has_engagement

Replies, reposts, quotes and threads

X search leaves native reposts out by default and keeps replies in, so the two most used type filters are -filter:replies and include:nativeretweets.

• filter:replies

  Finds

  Only replies

  Example

  from:nasa filter:replies

• -filter:replies

  Finds

  Only top-level posts

  Example

  -filter:replies from:nasa

• filter:self_threads

  Finds

  An author's replies to their own posts

  Example

  from:nasa filter:self_threads

• conversation_id:

  Finds

  Every post in one thread

  Example

  conversation_id:2110405289047048193

• filter:quote

  Finds

  Quote posts

  Example

  "launch" filter:quote

• quoted_tweet_id:

  Finds

  Quotes of one post

  Example

  quoted_tweet_id:2110405289047048193

• quoted_user_id:

  Finds

  Quotes of an account's posts, by numeric ID

  Example

  quoted_user_id:1592041833471827968

• include:nativeretweets

  Finds

  Reposts as well as posts

  Example

  from:nasa include:nativeretweets

• filter:nativeretweets

  Finds

  Only reposts made with the button

  Example

  from:nasa filter:nativeretweets

• filter:retweets

  Finds

  Old "RT" style retweets and quotes

  Example

  from:nasa filter:retweets

Media and links

Media filters match what is attached to a post; link filters match what it points to. Every card_name: operator only reaches back about 7 days, polls included: card_name:poll2choice_text_only finds two-option text polls.

• filter:media

  Finds

  Any image, video or GIF

  Example

  #nyc filter:media

• filter:images

  Finds

  Images, linked ones included

  Example

  #nyc filter:images

• filter:twimg

  Finds

  Images uploaded to X

  Example

  #nyc filter:twimg

• filter:videos

  Finds

  Any video, YouTube links included

  Example

  keynote filter:videos

• filter:native_video

  Finds

  Videos uploaded to X

  Example

  keynote filter:native_video

• card_name:animated_gif

  Finds

  GIFs

  Example

  monday card_name:animated_gif

• filter:spaces

  Finds

  Links to X Spaces

  Example

  ai filter:spaces

• filter:links

  Finds

  Any link, media links included

  Example

  -filter:media filter:links

• url:

  Finds

  Links to a domain; hyphens become underscores

  Example

  url:github.com

• filter:news

  Finds

  Links to news sites

  Example

  election filter:news

• filter:hashtags

  Finds

  Posts with any hashtag

  Example

  from:nasa filter:hashtags

• filter:mentions

  Finds

  Posts that mention any account

  Example

  from:nasa filter:mentions

• filter:safe

  Finds

  Leaves out posts marked sensitive

  Example

  anime filter:safe

Language and app

lang: takes a two-letter language code and trusts X's own guess, which is weakest on very short posts.

• lang:

  Finds

  Posts in one language

  Example

  ramen lang:ja

• lang:und

  Finds

  Posts X can't classify, such as emoji only

  Example

  lang:und filter:images

• lang:qme

  Finds

  Posts made of media links only

  Example

  lang:qme from:nasa

• source:

  Finds

  Posts sent from one app; underscores for spaces

  Example

  source:tweetdeck

Location

Location operators only see posts that carry a place or coordinates, which is a small share of what people post, so a location query returns a thin slice of the conversation.

• near:

  Finds

  Posts tagged near a place; quotes for two words

  Example

  coffee near:boston

• within:

  Finds

  A radius for near:, in km or mi

  Example

  coffee near:tokyo within:15km

• geocode:

  Finds

  Posts within a radius of coordinates

  Example

  geocode:40.7128,-74.0060,5km

• place:

  Finds

  Posts tagged with one X place ID

  Example

  place:96683cc9126741d1

Date and time operators come next, in a section of their own.

#How to search Twitter by date

To search X by date, add since: with the first day and until: with the day after the last one, both in YYYY-MM-DD form: since: includes its date, until: stops at the start of its date, and both days begin at midnight UTC.

1. Start with the topic or the account. "solar eclipse", or from:nasa for one account's posts.

2. Add the first day. since:2026-10-01 includes October 1.

3. Add the day after the last day. until:2026-11-01 covers all of October, because November 1 itself is left out.

4. Read the Latest tab. It lists every match in time order; Top shows a ranked selection.

The whole query reads from:nasa since:2026-10-01 until:2026-11-01. One day is since:2026-10-15 until:2026-10-16. On X's Advanced Search form, the From and To fields under Dates write the same two operators.

Good to know

Dates follow UTC, not your clock. The UTC day of October 15 runs from 8 p.m. on October 14 to 8 p.m. on October 15 in New York, so a search for that date catches the evening before and misses the evening itself. For a New York day, give the hours in UTC: since:2026-10-15_04:00:00_UTC until:2026-10-16_04:00:00_UTC.

Pick a window below and copy the operators in the form you need, for x.com, for Sorsa API or for the official X API.

Hours, Unix time and post IDs

For windows shorter than a day, X search takes times and post IDs as well as dates. Each line below is one operator with what it finds:

since:2026-10-01                 # on or after a UTC date
until:2026-11-01                 # before a UTC date
since:2026-10-15_14:00:00_UTC    # from a time of day, in UTC
until:2026-10-15_18:00:00_UTC    # before a time of day
since_time:1792072800            # from a Unix second
until_time:1792087200            # before a Unix second
within_time:2d                   # the last 2 days; also h, m and s
since_id:2108468174910472193     # newer than a post ID
max_id:2110405289047048193       # that post ID and older

Every post ID since November 2010 carries the moment it was created, which is why since_id: and max_id: work as precise time bounds. The free Tweet ID to Date Converter reads that time out of any ID. Date operators need at least one other term in the query to return anything.

How far back X search goes

X search reaches back to the first public post of March 2006, and the date form accepts any day since then. Old results thin out wherever posts were deleted or accounts went private, because X search only returns what is public today. To see what one account posted on a given day, use the free Search Tweets by Date tool; for years of results in bulk, our guide to historical Twitter data covers backfills.

#Search operators in the official X API

The official X API v2 reads its own dialect of these operators: it renames most filters, takes dates as request parameters and rejects the web names min_faves: and min_retweets: with a 400 error (X's operator reference (opens in a new tab)). Sorsa API reads the same operators as the search box on x.com, so a query moves between the two unchanged.

• At least 100 likes

  Sorsa API and x.com

  min_faves:100

  Official X API v2

  min_likes:100

• At least 20 reposts

  Sorsa API and x.com

  min_retweets:20

  Official X API v2

  min_reposts:20

• At least 10 replies

  Sorsa API and x.com

  min_replies:10

  Official X API v2

  min_replies:10

• October 2026 only

  Sorsa API and x.com

  since:2026-10-01 until:2026-11-01

  Official X API v2

  start_time, end_time parameters

• The last 2 days

  Sorsa API and x.com

  within_time:2d

  Official X API v2

  start_time parameter

• Only replies

  Sorsa API and x.com

  filter:replies

  Official X API v2

  is:reply

• Only quote posts

  Sorsa API and x.com

  filter:quote

  Official X API v2

  is:quote

• Quotes of one post

  Sorsa API and x.com

  quoted_tweet_id:

  Official X API v2

  quotes_of_tweet_id:

• Posts with images

  Sorsa API and x.com

  filter:images

  Official X API v2

  has:images

• Posts with any media

  Sorsa API and x.com

  filter:media

  Official X API v2

  has:media

• Posts with links

  Sorsa API and x.com

  filter:links

  Official X API v2

  has:links

• Verified authors

  Sorsa API and x.com

  filter:verified

  Official X API v2

  is:verified

• Reposts

  Sorsa API and x.com

  Left out by default

  Official X API v2

  Left out of keyword searches

• Accounts you follow

  Sorsa API and x.com

  x.com only, signed in

  Official X API v2

  No operator

• Words, from:, to:, @, #, $, lang:, list:, url:

  Sorsa API and x.com

  The same

  Official X API v2

  The same

Two changes in 2026 explain why older lists and AI answers disagree about the official API. X switched off min_likes:, min_replies: and min_reposts: on its search endpoints on January 19, 2026, and switched them back on May 4, when search moved to a new index that also stopped returning retweets for keyword queries (X's API changelog (opens in a new tab)). Claims that the official API "ignores" engagement filters describe the months in between; today it answers the web names with an error.

The official endpoints also cap the query string: 512 characters on recent search, which covers the last 7 days, and 1,024 on full-archive search. Conjunction-only operators such as has:media or is:reply can't make up a query on their own there; they need a word, an account or a hashtag beside them.

#Run the same queries in code with Sorsa API

For running X search operators from code, use Sorsa API, a Twitter (X) API and scraper: POST /search-tweets takes the query exactly as you would type it on x.com, returns about 20 tweets per request with each author's full profile, and costs from $0.07 per 1,000 results. The official X API charges $5.00 per 1,000 posts, needs a developer account and OAuth, and reads its own operator names instead of the x.com ones.

1. Get a key. Sign up and copy the API key from your dashboard. The first 100 requests are free.

2. Send the query. Put the operators in the query field of a JSON body; order picks popular (Top, the default) or latest.

   curl -X POST "https://api.sorsa.io/v3/search-tweets" \
     -H "ApiKey: YOUR_API_KEY" \
     -H "Content-Type: application/json" \
     -d '{
       "query": "\"vector database\" lang:en min_faves:50 since:2026-10-01",
       "order": "latest"
     }'
   

3. Read the tweets. Each one arrives with its counts, its date in ISO 8601 and its author's profile.

   {
     "tweets": [
       {
         "id": "2110405289047048193",
         "full_text": "Benchmarked three vector database engines on 40M vectors.",
         "created_at": "2026-10-14T16:20:05Z",
         "lang": "en",
         "likes_count": 184,
         "retweet_count": 41,
         "reply_count": 23,
         "quote_count": 6,
         "view_count": 52300,
         "is_reply": false,
         "user": {
           "id": "1592041833471827968",
           "username": "kestrel_db_engineering",
           "display_name": "Kestrel DB Engineering",
           "followers_count": 12480,
           "verified": false
         }
       }
     ],
     "next_cursor": "DAABCgABGSmiaxkAAgoAAgjEJ"
   }
   

   Show full response

4. Follow the cursor. Send next_cursor back with the same query until a response comes without one.

   import requests
   
   URL = "https://api.sorsa.io/v3/search-tweets"
   HEADERS = {"ApiKey": "YOUR_API_KEY"}
   
   
   def search(query, order="latest"):
       body = {"query": query, "order": order}
       while True:
           r = requests.post(URL, headers=HEADERS, json=body)
           r.raise_for_status()
           page = r.json()
           yield from page["tweets"]
           cursor = page.get("next_cursor")
           if not cursor:
               break
           body["next_cursor"] = cursor
   
   
   q = '"vector database" lang:en min_faves:50 since:2026-10-01'
   for tweet in search(q):
       print(tweet["created_at"], tweet["user"]["username"])
   

   const URL = "https://api.sorsa.io/v3/search-tweets";
   const headers = {
     ApiKey: "YOUR_API_KEY",
     "Content-Type": "application/json",
   };
   
   const body = {
     query: '"vector database" lang:en min_faves:50 since:2026-10-01',
     order: "latest",
   };
   do {
     const res = await fetch(URL, {
       method: "POST",
       headers,
       body: JSON.stringify(body),
     });
     const page = await res.json();
     for (const t of page.tweets) console.log(t.created_at, t.user.username);
     body.next_cursor = page.next_cursor;
   } while (body.next_cursor);
   

A 403 means the key is missing or its requests have run out, and a 429 means more than 20 requests in one second; failed requests are never charged. The POST /search-tweets reference (opens in a new tab) lists every field, and our Twitter search API guide covers polling a query on a schedule.

The cost of a search job follows the number of results. Sorsa API charges by the request, about 20 results each, while the official X API bills every post it returns:

• 1,000

  Sorsa API, from

  $0.07

  Official X API

  $5.00

• 100K

  Sorsa API, from

  $7

  Official X API

  $500

• 1M

  Sorsa API, from

  $70

  Official X API

  $5,000

Your 100 free requests cover about 2,000 search results, so a first test of any query above costs nothing.

#Ready-made search queries

These queries cover the searches people run most; each works in the search box on x.com and in the query field of Sorsa API as written. Swap in your own words, accounts and dates. On x.com, the address of a results page keeps the query, so a bookmark saves the search, and &f=live at the end of the address opens the Latest tab.

# What people say about a brand, without its own posts
("acme cloud" OR @acme_cloud_status) -from:acme_cloud_status -filter:replies

# Questions in a niche that nobody has answered yet
("any recommendations" OR "what do you use") crm ? -min_replies:1 lang:en

# The most liked posts on a topic this month
"vector database" min_faves:200 -filter:replies since:2026-10-01

# One account's posts on a subject in October 2026
from:kestrel_db_engineering benchmark since:2026-10-01 until:2026-11-01

# Complaints about a competitor
@brightpath_analytics (broken OR outage OR refund OR "not working")

# Every reply in one thread
conversation_id:2110405289047048193

# Quote posts of one post
quoted_tweet_id:2110405289047048193

# Images under a hashtag, original posts only
#buildinpublic filter:images -filter:replies

# Posts that link to your site, from other accounts
url:acmecloud.dev -from:acme_cloud_status

# Hiring posts with links, from the last 2 days
(hiring OR "we're hiring") "rust engineer" filter:links within_time:2d

To follow every post that tags one account, with like, reply and date filters in the request itself, see our guide to the Twitter mentions API.

#Why a search returns nothing

When a query comes back empty, the cause is usually in the syntax, not in X. Check these first:

• A space after a colon, a lowercase or, or a date not written as YYYY-MM-DD.
• The end date. until: leaves its own day out, so until:2026-10-31 misses October 31.
• The Top tab. Top ranks and trims results; switch to Latest, or send "order": "latest" to Sorsa API.
• Private accounts. Protected and suspended accounts never appear in search, for anyone.
• Short reach. card_name: covers about the last 7 days, and location filters only see tagged posts.
• Too many operators. Past about 22, split the query in two and merge the results.
• Hyphens in url:. Write url:t_mobile.com for t-mobile.com.
• Words X reads as filters. Searching photo or video can turn into a media filter; put the word in quotes.
• Official API names. has:media, is:reply and min_likes: belong to the official X API; on x.com and Sorsa API use filter:media, filter:replies and min_faves:.

Five pairs cause most of the wrong results:

• to: and @

  The difference

  to: finds replies to the account; @ finds any post that tags it

• filter:retweets and filter:nativeretweets

  The difference

  Old "RT" text retweets and quotes, against reposts made with the button

• filter:nativeretweets and include:nativeretweets

  The difference

  Only reposts, against reposts added to normal results

• filter:images and filter:twimg

  The difference

  Any image, linked ones included, against images uploaded to X

• min_faves: and min_likes:

  The difference

  The same floor: the x.com name, and the official X API's name

No operator lists the accounts that liked a post, and none searches private messages or protected accounts. Search covers what is public on X today.

#Search X with operators from your code

Twitter search operators turn a keyword search into a precise filter, and the same query that works on x.com runs in code through Sorsa API as written. Build the query in the search box or the Twitter Advanced Search tool, then send it to POST /search-tweets: about 20 tweets per request, any date back to 2006, 20 requests per second on every plan and from $0.07 per 1,000 results. Your first 100 requests are free.

#FAQ