Technology — a classification, not a measurement. An on-box language model (Qwen3.6-35B-A3B-UD-Q6_K_XL, prompt version 1) read this channel’s own recent posts on 11 August 2026 and assigned it the closest of 31 fixed categories, at 97% confidence. This is a model’s judgement about what the channel is likely to be about, not a fact this register measured the way a subscriber count or a view count is measured — it can be revised on a later pass, and it carries no weight anywhere else on this page. How this classification works, and why it has no browse page of its own yet.
Observations
These are measurements, not verdicts. Each one below states something we counted, alongside the evidence it was counted from, so you can check it rather than take it. None of them is graded: every observation this register holds is recorded at severity 0, because the precision of the detectors behind them has not been measured yet, and a rating we cannot support is worse than none. Read each as a fact about the data, not as a judgement about the channel. How we measure.
Content that also appears on other registered channels
Posts published here appear word for word on 1 other registered channel. The matching is on the text itself, not on Telegram’s forward marker, so it finds a copy whether or not it was labelled as one.
Matching posts — open both and compare (6 of the pairs behind the counts below)
Text overlap is the Jaccard coefficient over the set of distinct three-word phrases in the two bodies: 1.00 is identical wording, and the threshold for counting a pair at all is 0.70. Candidates are generated by simhash LSH (4 x 16-bit bands, exact Hamming <= 3) verified against the bodies with Jaccard over the SET of distinct 3-word shingles. Published first counts which side of each matching pair carries the earlier timestamp — in this corpus, which is the limitation directly below.
What this cannot establish
MEASURED, DOMINANT ERROR SOURCE: a post ingested before 2026-08-06 may have carried a forward header that was not recorded. A 45-pair hand-check against live t.me pages found 14 (31%) where the live page shows a forward header naming the other channel and the database has none, plus 4 more (9%) naming a third party. The text match itself was wrong 0 times out of 45. Read attribution_capture.items_in_trusted_window before treating the unattributed count as a claim.
Telegram lets a channel forward a post with a header naming the source, and we only began reliably recording that header on 2026-08-06. None of the 1 matches recorded here fall after that date, so for this entry we cannot say whether any of them carried a credit. The duplication is measured; the absence of attribution is not.
“Published first” means first in this corpus. We hold 17 comparable posts for this entry, running 31 July 2026 to 7 August 2026. A channel we have read one page deep will look younger than a neighbour we have read in full, and the order would flip with no change in the underlying facts.
The detector’s own notes on this observation, as it recorded them. Names in this_style are fields of the underlying evidence record, which the plain-English paragraphs above read out for this entry.
Verbatim republication has three causes and the text separates only two: a clone/mirror, unattributed copy-paste, or BOTH channels copying a common third source that neither attributes. The spread filter (content held by at most 8 channels) reduces the third and does not remove it.
'Earliest' means earliest IN THIS CORPUS. A channel ingested one page deep will look younger than a neighbour ingested in full; corpus_coverage above is there to be checked before the direction is believed.
shared_verified_est extrapolates the sampled pass rate over the full narrow match count; sampled/passed are the numbers actually measured.
Absence of a forward header is not proof of intent: Telegram lets a channel disable forward attribution, and a credit written in the body is not parsed as attribution here (mention_edge_either_way above is the closest available signal).
Across the whole group of 2, the earliest publisher we hold is @AIKuhnya — which is this entry. That is a statement about our reading window, not a claim of authorship.
Recorded under the key clone_source, last confirmed 7 August 2026. An observation that a later pass no longer finds is cleared, and a cleared observation is removed from this page rather than being shown struck through — we do not keep publishing a claim we have withdrawn. Dispute an observation.
Also posting the same content
This channel’s posts match, word for word or near enough, posts on 1 other registered channel, found by comparing text fingerprints across every channel on the register. That matching has been checked by hand against the live Telegram pages and found reliable — 0 wrong of 45 pairs re-read.
Which channel, if either, published first is deliberately not shown. The same hand-check found that reading wrong 18 of 45 times — 60%, no better than a coin flip — because it depends on how deep our own crawl happened to reach into each channel’s history, not on when the content was actually first posted. This list is ordered by subscriber count, the same as every other listing on this site, never by which channel we think came first. Word-for-word matching has several ordinary explanations besides copying — a channel mirroring itself, an unattributed repost, or two channels independently repeating the same wire story — and this measurement cannot tell those apart. How this is measured.
3 measurements spanning 3 days, net +22. Dots are measurements; the straight line between them is drawn to join them, not to claim we know the path taken in between — snapshots are recorded only when a count changes, so gaps mean “no change observed”, never “interpolated”. The vertical axis spans 4,031–4,059 and does not start at zero.
Measurement log — every subscribers count we have recorded
Measured (UTC)
Subscribers
Change
11 Aug 2026, 00:16
4,056
+22
8 Aug 2026, 03:30
4,034
no change
7 Aug 2026, 19:18
4,034
first reading
Engagement
20 posts held, back to 31 July 2026 — the reader has not yet reached the start of this channel’s public history, so older posts may sit further back, unread. Read across 2 pagesof Telegram’s post history, 20 posts per page.
ERR · 30 days
13.3%
avg views ÷ 4,056 subscribers
Avg views / post
537
20 posts measured
Reaction rate
0.586%
reactions ÷ views · ER floor
Posts in window
20
of 20 held
ERR is average views per post over the last 30 days divided by subscribers, the definition TGStat uses, so this figure is comparable with the one you will see elsewhere. It falls structurally as a channel grows: a high ERR on a small channel and a low one on a large channel describe reach mathematics, not quality. We publish the figure and the sample it came from and pass no verdict on it.
ER is defined industry-wide as (forwards + reactions + comments) ÷ views— note the denominator is views, not subscribers. Telegram’s public web preview carries views and reactions but not forward or comment counts, so the reaction rate above is the reactions term only and is therefore a floor: the true ER for this channel is higher by an amount we have not measured and will not estimate.
What these figures were computed from
Window
Rolling 30 days · latest post in window 7 August 2026
Posts held
20 (31 July 2026 – 7 August 2026)
Views total
10,749
Reactions total
63
Forwards / comments
not exposed by the public surface — not measured, not estimated
Readings taken
8 Aug 2026, 05:42 UTC
Views are a single reading per post, taken at the time above. A post published in the last day or two is still accumulating views, which pulls the 30-day average down slightly. That is a property of the standard definition rather than a fault in it, so we keep the definition rather than “correcting” the number into something nobody can reproduce.
Precision. Telegram publishes view counts on its public widget in short form — 8.12K, 3.7M — so any reading at or above 1,000 reaches us rounded to three significant figures, and only counts below 1,000 are exact. Averages and rates derived from them are shown to the same precision rather than to the unit: a figure like 3,701,250 would assert digits nobody measured.
Reaction counts are published per emoji and rounded the same way, so a total below 1,000 is exact and a larger one is a sum that may carry a rounded component from each emoji above 1,000. Because it is a sum, it does not look rounded — read a large reaction total as three significant figures per contributing emoji rather than as the figure it prints.
What this channel posts
Video runtime
34m 00s
Average length
4m 15s
Measured directly from 8 videos with a duration reading, out of the posts we hold for this channel — not this channel’s whole posting history, only the sample this register has actually read. An exact reading to the second, taken from the post itself rather than from Telegram’s own rounded chrome, so it carries no ≈ mark.
Reaction mix
63 reactions across 20 posts, in 4 distinct kinds. The most used accounts for 58.7% of them.
Every reaction kind recorded on the sample, most used first
Reaction
Count
Share
Share, drawn
❤
37
58.7%
👍
18
28.6%
😁
6
9.52%
🔥
2
3.17%
No sentiment is inferred, and none should be read in. This table is ordered by count and by nothing else. Emoji do not carry stable meaning across languages or communities — 🙏 is thanks in one channel and mourning in another — so we publish which ones were pressed and how often, and pass no judgement on what an audience meant by them.
Precision. Telegram publishes reaction counts per emoji and short-forms each one — 4.34K, 1.2M — so any single kind at or above 1,000 reaches us at three significant figures, and only counts below 1,000 are exact. The shares above are ratios of those figures and carry the same error. This is also why the total here can differ slightly from a reaction total printed elsewhere on the page: both are sums of the same rounded parts, taken over samples with different edges.
Coverage. Reactions were read on 20 of the 20 sampled posts in this sample. Summed by Telegram’s own count on each post — not by adding up the per-emoji breakdown above — those same posts carry 63reactions in total: the kind of figure the paragraph above means by “a reaction total printed elsewhere on the page”.
Measured over the 20 most recent posts we hold, published 31 July 2026 to 7 August 2026, using the newest reading held for each. Telegram Stars are excluded: they are a payment, not a reaction, and they have their own section.
АЙТИШНИКИ БЕСПЛАТНОЕ ОБУЧЕНИЕ сборник курсов, инструментов и книг
Проект «TERMINAL» стал крупнейшей библиотекой бесплатного образования. В одном канале собраны курсы, книги, полезные инструменты и практические тренажёры для всех разработчиков
🎓 Практические курсы и задания
🪽 Книги и статьи известных авторов
😮💨 Полезные инструменты и ресурсы
🌟 IT-новости и инсайды
Обучение по всем направлениям: SQL, Python, Fro…
Meta выпустила альтернативу Claude Code
Новый инструмент называется Muse Code и работает на модели Muse Spark 1.2. Из коробки доступны скиллы и субагенты для автоматизации разработки.
По стоимости API оказался дешевле DeepSeek V4 Flash, но есть нюанс: низкие цены доступны только тем, кто разрешает Meta использовать свой код для обучения модели.
Подписки у сервиса нет, Muse Code работает исключительно через API. Ус…
Превращаем ИИ-агента в реверс-инженера
Нашли Reverse Skill для Claude Code, Codex и Cursor. Достаточно загрузить APK-файл, сайт или программу, после чего агент сам подберёт нужные инструменты, разберёт устройство приложения и проанализирует, как работают его функции.
Полученные механики можно использовать как основу для реализации похожего функционала в собственных проектах.
Забираем здесь.
ИИчница | ИИ-кухня
Программисты, это вам 👇
Держите 5 каналов, которые реально помогают изучать программирование и IT с полного нуля:
🖥 Easy Coder — все направления IT.
👩💻 Easy Python — всё о Python.
🌐 Easy WebDev — Frontend, Backend.
🔠 Easy InfoSec — ИБ, Хакинг.
🖥 Easy GitHub — лучшее с GitHub.
Тонны бесплатной инфы для любого уровня подготовки ✔️
Локальный ИИ-коллега от основателя Coursera уже на GitHub
OpenWorker умеет работать с терминалом, файлами и популярными сервисами вроде GitHub, Jira и Slack.
Запускается на Windows и macOS, а в качестве мозга поддерживает OpenAI, Anthropic, Gemini и локальные модели через Ollama.
Забираем тут.
ИИчница | ИИ-кухня
На Stepik вышел курс «MCP: создание передовых агентных приложений»
Практическай материал по разработке AI-агентов и многосерверных систем на базе Model Context Protocol
Шаг за шагом собираем своего AI-агента с серверами, инструментами, логикой для принятия решений и даже анализом изображений. Всё это на практических примерах, с схемами и кодом в терминале
Что внутри:
✅ клиент-серверная архитектура MCP
✅ серверные…
Совет по вайбкодингу: подключите GitHub к своему ИИ-агенту и не просите сразу писать код. Это заметно сократит расход токенов.
Новички начинают с запроса вроде: «Напиши мне приложение, которое принесёт $100000, не допускай ошибок». Но гораздо эффективнее сначала дать агенту изучить существующие опенсорс-проекты.
Промпт:
«Я разрабатываю XXX. Пока не создавай файлы и не пиши код. Сначала найди похожие open source-про…
Вышла Seedance 2.5 — видеомодель получила крупное обновление с упором на контроль генерации и длинные ролики.
Как и обещали ранее, теперь модель умеет создавать 30-секундные видео и поддерживает консистентное продолжение до 180 секунд в бета-режиме.
На вход можно подать до 50 референсов. Разработчики сделали ставку на профессиональные пайплайны: появилась оптимизация для работы с гринскрином и 3D-болванками из Blen…
Вышел Qwen 3.8-Max
Это модель на 2,4 трлн параметров с 95 млрд активных. По бенчмаркам она конкурирует с Sol и Fable. Немного уступает моделям Anthropic только на SWE Pro, зато показывает отличные результаты на TerminalBench.
Разработчики утверждают, что модель способна автономно работать до 16 дней. В качестве примера они показали, как агент за пару недель без вмешательства человека с нуля написал небольшую CLI-ут…
❤5👍1
Showing the 12 most recent of 20 posts we hold for @AIKuhnya. View and reaction counts are the latest single reading for each post, not a live figure, and a recent post is still accumulating both. A view count marked ≈ was rounded by Telegram before we ever saw it — t.me prints views in full below 1,000 and to three significant figures above, so ≈1,200,000 means somewhere between 1,150,000 and 1,249,999. Unmarked counts are exact. Text is reproduced from the public post preview and truncated for length.
Mentions
Names
Channels on the register whose handles appear in this channel's posts.
A mention is a weaker signal than a forward and is counted separately for that reason — naming a channel is not republishing it, and a handle in a post body is easy to place deliberately. The post counts beside each row below are distinct posts in which the handle appeared, from posts we have read on both sides — the “Named by N registered channels” figure above is a different count, of distinct NAMING CHANNELS rather than posts, and is not the sum of the rows under it.
Cite this entry
A live page changes as we take new readings, so a citation should name the measurement it is based on, not just the URL. The line below cites the subscriber count as measured 11 August 2026 — this
entry's latest reading, not the date you are reading this.
“ИИчница | ИИ-кухня” (@AIKuhnya), 4,056 subscribers as measured 11 August 2026. Telegram Register, tgregister.com/channel/AIKuhnya.
Full measurement history, CC BY 4.0. Every reading this register holds for this entry, not just the latest one, as a dated, downloadable record: CSV · JSON. Free to use with attribution to tgregister.com. Each file carries its own generation timestamp, which is the figure to cite for exactly when the data was retrieved.
