The community can be activated on demand
MEASURED#ori, 2026-08-21: 200 × $5 coupons cleared in about a minute. That
single day produced 308 of the channel’s 452 lifetime messages (68%) from 73 of its 116
lifetime authors (63%). Volume fell to 6 messages/day within 24 hours, and 2/day by 08-25.
INFERRED Attention is acquirable on demand. Nothing currently converts a spike into retained usage — the measurement that would tell you whether that spend was acquisition or leakage was available and never made.
Two support channels failing in opposite directions
MEASURED On #community-help, an assistant replies in 0 minutes, 10 of 10 sampled
threads — but 8 of those 10 never saw a human, and across aged threads roughly 6 in 10 got no
human reply at all. Nobody is being left waiting on that channel, which is the harder half
to build and it is already built. A fast-answering community with a measurement blind spot is a
different thing from a neglected one, and this is the former.
#providers corroborates outside that census: 52 threads, 45 of them in
its launch month, then roughly twelve months of nothing, still in the sidebar.
INFERRED Because the two channels fail in opposite directions, no single response-rate metric can see either. A dashboard built on either one alone would report this community as healthy. That is the measurement gap, and it is the same gap that makes paid activation unaccountable.
Usage proof already exists and isn’t crawlable
This is the cheapest high-value item I found anywhere.
MEASURED/openai/gpt-4o/apps, /activity, /providers and /uptime all
return HTTP 200 with distinct content. Zero of them appear in sitemap.xml, at any depth,
for any author.
INFERRED So “which apps actually use this model” — real usage proof, the exact outcome signal a comparison table can’t carry — is already built and simply not exposed. That reframes it from a publishing programme into a sitemap change plus a template decision.
What you publish for machines is advanced
MEASURED Fetched 2026-08-27, public endpoints, no authentication:| Document | Size / count |
|---|---|
openrouter.ai/llms.txt | 72,876 B |
openrouter.ai/llms-full.txt | 3,642,174 B |
openrouter.ai/docs/llms.txt | 69,424 B |
openrouter.ai/{author}/{model}/llms.txt | ~4,066 B per model |
sitemap.xml | 4,980 URLs |
robots.txt | Allow: /; only /seo/ disallowed |
Per-model llms.txt at catalogue scale is ahead of most infrastructure companies. Any proposal
that opens by telling this team to think about answer-engine optimisation was written by someone
who didn’t look. You thought about it first.
The site’s content is product-generated
MEASURED First-path-segment tally of all 4,980 sitemap URLs:/compare 3,051
(61.3%), /docs 405, model pages ~550, /blog 120, /provider 105, /spawn 91,
/works-with-openrouter 54, /apps 33.
MEASURED I opened the slugs rather than trusting the segment names, which changed
the reading twice. /apps is a category taxonomy, not app pages. /works-with-openrouter is
a real third-party integration directory — aider, browser-use, boltai, agent-zero, Maxim AI.
/spawn is a product page. /blog resolves to tutorials, announcements, insights —
there is no case-study section.
INFERRED The gap is narrower and more specific than a raw tally suggests: 54 listings that a tool integrates, and no page describing what anyone achieved by integrating. No workload, no cost delta, no routing decision explained. The missing class isn’t third-party presence — it’s outcome.
Presence outperforms announcement
MEASURED In#announcements, normalised within year — reaction counts inflate
over time, with yearly medians of 11 / 18 / 41 / 50 for 2023→2026, so the all-time median of 25
is an invalid yardstick — advance event notices score 9–28 reactions. Posts announcing that
something is happening right now score 92 and 101.
MEASURED #events holds six messages in its entire life, 13 reactions total,
two events at zero, no threads, no recaps.
MEASURED Meanwhile lu.ma/openrouter
carries 12 events across SF, NYC, London and Zoom.
Conference spend buys the bottom tier
MEASURED At AI Engineer World’s Fair 2026, OpenRouter bought Supporting Partner — the bottom tier, no booth — while four of its own six September hackathon co-partners bought higher tiers at the same event.The creator channel already has one worked example
MEASURED Fetched 2026-08-27: YouTube videoBOCU-seUXQ8, Build better e-commerce
websites · Web Dev Challenge S2.E12, owned by CodeTV, not by OpenRouter — published 2025-12-09,
35:50, with 8 mentions of OpenRouter in its description.
QUOTED The description opens “Sponsored by OpenRouter.” Under a Tool: OpenRouter
heading, the challenge brief reads “Your app must use OpenRouter (https://openrouter.ai/) as part of
the build.”
INFERRED That is a bought placement structured as a build requirement, not an ad read — the disclosure and the mandate sit in the same description, and the episode ships the repo where the build happened. It is the difference between renting attention and renting usage, and it is the placement the creator line in my plan budgets for.
How many such placements exist, I could not count. This one is visible because OpenRouter featured it on its own channel, so the method finds only what the company chose to surface; a census of CodeTV’s own feed returned nothing countable. At least one is the honest form.
Two founder interviews are not yet a channel
MEASURED Fetched 2026-08-27: OpenRouter’s founder appears on two podcasts six months apart — Around the Prompt (2025-02-17, 44:34) and Limitless (2025-08-04, 69:30). One of Around the Prompt’s 20 episodes mentions OpenRouter or its founder. The Limitless episode carries 3,145 views on YouTube, the only one of the two that publishes a counter. QUOTED Both are editorial bookings, not placements: “An unfiltered conversation with Alex Atallah, CEO of OpenRouter.”INFERRED That is what this channel looks like unbought — episodic, unrepeatable, and still in the low thousands a year after publication. A booking cannot be scheduled against a quarter, and two of them in a year is not reach. The case for buying podcast and newsletter placements is not that the audience is unproven; it is that earned appearances have already proven the audience and cannot be made to happen again on purpose.
Twenty-three long-form videos, no short-form
MEASURED Fetched 2026-08-27, counting grid items on three tabs of the same channel by the same method:/videos renders 23, /shorts renders 0, /streams renders 0.
The count on /videos is what makes the two zeroes readable — the same pattern that finds 23 there finds
none on the other two, and both empty tabs return the same fingerprint as each other.
openrouter.ai links four social accounts from its homepage: X, Twitter,
LinkedIn and YouTube. Instagram and TikTok are not among them — though both handles return a
generic shell to an unauthenticated fetch, so whether an unlinked account exists is not something
that page can tell me.
INFERRED The raw material is already made and already paid for: interview episodes running 30 to 40 minutes, self-hosted webinars, and two podcast appearances at 44 and 69 minutes. None of it has been cut down. That is not an absence of content — it is an absence of reformatting, which is the cheapest item on this list.
The posting speaks to enterprises; the growth it describes is developer self-serve
MEASURED Across all 853 announcement posts (2023-04-04 → 2026-08-26),"developers" falls from 8.4 to 1.3 mentions per 100 posts; "enterprise" first appears
2026-04-26.
QUOTED The posting describes OpenRouter as “the open AI routing and infrastructure
layer that enterprises use” — then, two paragraphs later, locates growth in developers finding it,
“nearly all of it starting in self-serve.”
PLATFORM FACT Discord messages sit behind authentication: not crawlable, not indexed. Whatever the community produces stays inside that boundary unless something carries it out.