Weekly release 2026-W17 · posts from 20–26 Apr 2026 · 97 readings
In 20 to 26 April 2026, Claude drew more people than Claude Code on Hacker News and a wide gap more than ChatGPT on Reddit; the figures count people posting, not their views.
Published 2 Oct 2026. Its figures never change. Cite it with its permanent link. What these states mean
March eraRebuilt from archives, with live capture from 7 Mar to 31 May.
The Hacker News summary table held subjects, and Claude drew people posting about it against for Claude Code. On Reddit, Claude drew people posting about it against for ChatGPT, a gap wide enough to set Claude's attention clearly apart. On the two platforms that have tables this week, then, attention gathered around Claude, which shows where people were posting and says nothing about what they made of it.
Under a published prompt · prose-ec-1.4 · sonnet-5.5 (agent setting) · release · run 540. How runs are published
The first place on each of the two platforms that count people, in its reading’s own words. Two platforms, two rankings; nothing here is added across them.
14,444 distinct people on Reddit
On Reddit, 14,444 distinct people posted about Claude between 20 and 26 April 2026: 23,188 posts, an estimated 8.00% of that week's Reddit posts about AI, leaving out templated automated posts (99.9% of these posts had an identifiable author).
Claude · Reddit · posts from 20–26 Apr 2026 · n 23,188 posts · author coverage 99.9% · 7 days: 7 backfilled
55 distinct people on Hacker News
On Hacker News, 55 distinct people posted about Claude between 20 and 26 April 2026: 56 posts, 16.72% of that week's Hacker News posts about AI (100.0% of these posts had an identifiable author).
Claude · Hacker News · posts from 20–26 Apr 2026 · n 56 posts · author coverage 100.0% · 7 days: 7 backfilled
Each card is the lead measured statement of one platform’s first place: the stored statement whose figures include that platform’s people count, shown verbatim. Platforms rank separately; a subject leading two platforms appears twice.
This release’s conditions
Every subject’s place on each platform that counts people, side by side. Read down a column for one platform’s ranking; read across a row to see how differently the platforms rank one subject.
| Rank | Subject · lineage | Reddit ▼ sortedn 289,735 postsbar: people, #1 = 14,444 | Hacker Newsn 335 postsbar: people, #1 = 55 |
|---|---|---|---|
| 1 | ClaudeAnthropic | 14,444#1 | 55#1 |
| 2 | ChatGPTOpenAI | 6,353#2 | Hidden · n 14 people: n 14 people, below the threshold of 20 people. |
| 3 | GeminiGoogle DeepMind | 4,000#3 | Hidden · n 5 people: n 5 people, below the threshold of 20 people. |
| 4 | Claude CodeAnthropic › Claude | 3,706#4 | 24#2 |
| 5 | GPTOpenAI | 3,411#5 | Hidden · n 9 people: n 9 people, below the threshold of 20 people. |
| 6 | QwenAlibaba | 2,625#6 | Hidden · n 7 people: n 7 people, below the threshold of 20 people. |
| 7 | GrokxAI | 1,979#7 | Hidden · n 0 people: n 0 people, below the threshold of 20 people. |
| 8 | CodexOpenAI | 1,968#8 | Hidden · n 17 people: n 17 people, below the threshold of 20 people. |
| 9 | DeepSeekDeepSeek | 1,510#9 | Hidden · n 6 people: n 6 people, below the threshold of 20 people. |
| 10 | GemmaGoogle DeepMind | 1,070#10 | Hidden · n 0 people: n 0 people, below the threshold of 20 people. |
| 11 | Microsoft CopilotMicrosoft AI | 1,027#11 | Hidden · n 1 people: n 1 people, below the threshold of 20 people. |
| 12 | KimiMoonshot AI | 878#12 | Hidden · n 5 people: n 5 people, below the threshold of 20 people. |
| 13 | GLMZhipu AI | 619#13 | Hidden · n 0 people: n 0 people, below the threshold of 20 people. |
| 14 | SoraOpenAI | 612#14 | Hidden · n 0 people: n 0 people, below the threshold of 20 people. |
| 15 | GitHub CopilotMicrosoft AI | 536#15 | Hidden · n 3 people: n 3 people, below the threshold of 20 people. |
People on Reddit · n 289,735 posts#1 = 14,444
People on Hacker News · n 335 posts#1 = 55
Distinct people who posted about each subject, 20–26 Apr 2026, as of 25 Sep 2026 19:30Z; published. Each column is scaled to its own #1 and ranks on its own; nothing is added across platforms, because one person can post on several. Rows follow Reddit, the default platform. Hatched cells are hidden, never shown as a number. The stored reason, e.g. Hacker News ChatGPT: “n 14 people, below the threshold of 20 people”. Each column’s n 289,735 posts on Reddit and 335 on Hacker News.
Posts, no peoplearXiv, Google News and Exa are not ranked. Article sources have no authors, so people aren’t counted. Posts per subject, largest first:
arXiv n 408 posts · 85 subjects
Qwen 14, GPT 13, Gemini 10, Claude 8 posts, and 81 more subjects
Google News n not recorded · 85 subjects
Not counted this week: Google News has no day between 20 and 26 April 2026 that AIDRAN counts, so there’s no figure to show.
Exa n not recorded · 85 subjects
Not counted this week: Exa has no day between 20 and 26 April 2026 that AIDRAN counts, so there’s no figure to show.
The same subjects one week later, on the platform with enough people. DeepSeek grew on Reddit while Claude eased.
Reddit · people per subject
Distinct people per subject on each release’s board, for subjects with n ≥ 100 posts in both weeks (31 on Reddit), as of 25 Sep 2026 19:30Z; published. Each panel on its own scale; lines join one subject’s two weeks. Each week’s n is its platform’s posts: Reddit n 289,735 and 289,624 posts. 2026-W17 and 2026-W18 follow on from each other, so their lines are joined.
When a post was written and when AIDRAN caught it are different days. For each platform, what was published each day against what was captured that day, and how much of the capture was catching up.What was published each day against what AIDRAN captured that day, and how much of it was catching up.
22 Apr: published 149,874 · captured 3,010
own scale, top = 150,221
24 Apr: published 6,523 · captured 6,776
own scale, top = 7,104
21 Apr: published 392 · captured 400 · late 2
own scale, top = 400
25 Apr: published 156 · captured 29
own scale, top = 187
21 Apr: published 396 · captured 151
own scale, top = 571
26 Apr: published 34 · captured 1,128 · late 1,100 · late heavy
own scale, top = 1,128
No capture this week: collection hadn’t started. Posts published on these days were counted from later captures.
own scale, top = 84
Per platform and day, 20–26 Apr 2026: posts published that day (published_n), posts captured that day whatever their publish date (captured_n), and the part of those captured late (late_captured_n), from conditions.coverage, computed 2 Oct 2026 07:23Z. Each platform on its own scale. A flag bar marks a partial, low-capture or late-heavy day. Not-started days show published only: those posts were captured later and do count.20–26 Apr 2026, from conditions.coverage, computed 2 Oct 2026 07:23Z. Each platform on its own scale; a flag bar marks a partial, low-capture or late-heavy day.
Families and their products on Reddit. A post about Claude Code usually also counts for Claude, so the bars sit together and are never summed.Families and their products on Reddit, side by side and never summed.
The family entry already includes its products; not added.
› Gemini API (Google AI Studio) 266 · Gemini CLI 104 · Gemini app 79
› Qwen Code 54
› Grok app 125
› DeepSeek API 38 · DeepSeek app 34
› Sora app 21
People on Reddit, 20–26 Apr 2026, as of 25 Sep 2026 19:30Z, all on Reddit’s one scale (#1 = 14,444). Groups follow each subject’s stored lineage.
Posts, threads and distinct people on each platform: the context for every share in this release. Each platform stands alone.The context for every share. Each platform stands alone.
| Platform | Posts | Threads | People | Author cov. |
|---|---|---|---|---|
| hidden | hidden | |||
| Bluesky | hidden | hidden | hidden | |
| YouTube | hidden | hidden | hidden | |
| Hacker News | 100.0% | |||
| arXiv | not applicable | 0.0% | ||
| Google News | hidden | hidden | not applicable | |
| Exa | hidden | hidden | not applicable |
20–26 Apr 2026 · as of 25 Sep 2026 19:30Z · counted from posts’ own data · no platform coverage-limited. People count distinct platform identities and are never added across platforms, so there is no total row. People are not applicable on article sources. Reddit’s posts include its comments. Each figure’s n 289,735 posts on Reddit, not recorded on Bluesky, not recorded on YouTube, 335 on Hacker News, 408 on arXiv, not recorded on Google News and not recorded on Exa.
Each subject against its previous stored reading under the same method, share and people, platform by platform. Each has its own page, for its size: one cell a subject and platform.
What AIDRAN was collecting on each platform, each day of the week. A day can hold posts even when collection hadn’t started: they were captured later.What AIDRAN was collecting on each platform, each day.
| Platform | Mon 20 | Tue 21 | Wed 22 | Thu 23 | Fri 24 | Sat 25 | Sun 26 |
|---|---|---|---|---|---|---|---|
| Mon 20 Apr: backfilled | Tue 21 Apr: backfilled | Wed 22 Apr: backfilled | Thu 23 Apr: backfilled | Fri 24 Apr: backfilled | Sat 25 Apr: backfilled | Sun 26 Apr: backfilled | |
| Bluesky | Mon 20 Apr: earlier method | Tue 21 Apr: earlier method | Wed 22 Apr: earlier method | Thu 23 Apr: earlier method | Fri 24 Apr: earlier method | Sat 25 Apr: earlier method | Sun 26 Apr: earlier method |
| YouTube | low captureMon 20 Apr: earlier method, low capture | Tue 21 Apr: earlier method | low captureWed 22 Apr: earlier method, low capture | low captureThu 23 Apr: earlier method, low capture | low captureFri 24 Apr: earlier method, low capture | low captureSat 25 Apr: earlier method, low capture | low captureSun 26 Apr: earlier method, low capture |
| Hacker News | Mon 20 Apr: backfilled | Tue 21 Apr: backfilled | Wed 22 Apr: backfilled | Thu 23 Apr: backfilled | Fri 24 Apr: backfilled | Sat 25 Apr: backfilled | Sun 26 Apr: backfilled |
| arXiv | late heavyMon 20 Apr: backfilled, late heavy | Tue 21 Apr: backfilled | Wed 22 Apr: backfilled | Thu 23 Apr: backfilled | Fri 24 Apr: backfilled | Sat 25 Apr: backfilled | Sun 26 Apr: backfilled |
| Google News | low capturelate heavyMon 20 Apr: earlier method, low capture, late heavy | low capturelate heavyTue 21 Apr: earlier method, low capture, late heavy | low capturelate heavyWed 22 Apr: earlier method, low capture, late heavy | late heavyThu 23 Apr: earlier method, late heavy | late heavyFri 24 Apr: earlier method, late heavy | late heavySat 25 Apr: earlier method, late heavy | late heavySun 26 Apr: earlier method, late heavy |
| Exa | Mon 20 Apr: not started | Tue 21 Apr: not started | Wed 22 Apr: not started | Thu 23 Apr: not started | Fri 24 Apr: not started | Sat 25 Apr: not started | Sun 26 Apr: not started |
Coverage computed 2 Oct 2026 07:23Z · as of 25 Sep 2026 19:30Z · frozen with the release, never recomputed. Each day opens what it holds.A bar marks a flagged day.
Published 2 Oct 2026. Its figures never change. Cite it with its permanent link.
AIDRAN. Release 2026-W17.
Permanent link /releases/2026-W17
AIDRAN is the author and the publisher, so the citation names it once. To cite one subject, cite its reading page instead.
Licensed CC BY 4.0: reuse it with this citation.
None. No reading in this release has been revised or corrected.