Why is it possible to get information only about the first 9249 chat users?

Question

API

I tried to collect all the participants of a large chat (30,000 people) with various libraries,
but all libraries give out only 9249 participants. Somehow this is the limit.
Does anyone have any guesses why?

Code example

import sys
from getpass import getpass
from time import sleep

from telethon import TelegramClient
from telethon.errors import SessionPasswordNeededError
from telethon.errors.rpc_errors_400 import UsernameNotOccupiedError
from telethon.errors.rpc_errors_420 import FloodWaitError
from telethon.tl.functions.channels import GetParticipantsRequest
from telethon.tl.functions.contacts import ResolveUsernameRequest
from telethon.tl.types import ChannelParticipantsSearch, InputChannel

# First you need create app on https://my.telegram.org
api_id = 0
api_hash = ''
phone = ''
limit = 100


def get_chat_info(username, client):
    try:
        chat = client(ResolveUsernameRequest(username))
    except UsernameNotOccupiedError:
        print('Chat/channel not found!')
        sys.exit()
    result = {
        'chat_id': chat.peer.channel_id,
        'access_hash': chat.chats[0].access_hash
    }
    return result


def dump_users(chat, client):
    counter = 0
    offset = 0
    chat_object = InputChannel(chat['chat_id'], chat['access_hash'])
    all_participants = []
    print('Process...')
    while True:
        participants = client.invoke(GetParticipantsRequest(
                    chat_object, ChannelParticipantsSearch(''), offset, limit
                ))
        if not participants.users:
            break
        all_participants.extend(['{} {}'.format(x.id, x.username)
                           for x in participants.users])
        users_count = len(participants.users)
        offset += users_count
        counter += users_count
        print('{} users collected'.format(counter))
        sleep(2)
    with open('users.txt', 'w') as file:
        file.write('\n'.join(map(str, all_participants)))


def main():
    channel_name = input('Input a channel name, without "@": ')
    client = TelegramClient('current-session', api_id, api_hash)
    print('Connecting...')
    client.connect()
    if not client.is_user_authorized():
        try:
            client.send_code_request(phone)
            print('Sending a code...')
            client.sign_in(phone, code=input('Enter code: '))
            print('Successfully!')
        except FloodWaitError as FloodError:
            print('Flood wait: {}.'.format(FloodError))
            sys.exit()
        except SessionPasswordNeededError:
            client.sign_in(password=getpass('Enter password: '))
            print('Successfully!')
    dump_users(get_chat_info(channel_name, client), client)
    print('Done!')

if __name__ == '__main__':
    main()

Works great with small chats. But when trying to parse chat users of 30,000 people, it parses only the first 10,000 (9249, to be more precise).
Tried other libraries - all up to 10,000 approximately work