LLM Gateway / field notes
Marketing retrospective / 27 September 2026

The product became
the marketing system.

Four audiences. A substantial marketing foundation. Now checked against seventeen months of PostHog data, from May 2025 to today, with financial information excluded.

Source-backed, now measured. Marketing visible in the repository through 27 September 2026, supplemented by authorized non-financial observations from Search Console, PostHog and the admin Audience panel. Not a complete account of off-repository campaigns or causal business outcomes. Observed activity is not proof of causal lift. Financial metrics and personal data are excluded. The homepage refresh and shared claims were concurrent local work; implementation is not proof of deployment.
A / May 2025 to now

Seventeen months.
Search did the compounding.

Full PostHog history, queried through the PostHog MCP, from the first tracked day to today. People are counted per month. Financial data is excluded.

Window. 2025-05-01 to 2026-09-28 (exclusive), UTC. May 2025 is the first partial month of tracking (from May 30). September 2026 covers the 1st to 27th only. Partial months are drawn in rust. The weekly detail from the earlier refresh follows in section B.

People in the month

37KSep 2026 (27 days)

June 2025: 1,522
PostHog persons with a pageview

People from Google

40.1KQ3 2026 to date

Q3 2025: 2,985
Pageview referrer, not first touch

People creating an API key

1.4KSep 2026 (27 days)

June 2025: 17
Includes existing users

People from AI assistants

450Sep 2026 (27 days)

June 2025: 10
Referrer only; UTM-tagged visits excluded

Monthly people across every product site

Two step changes: the July 2025 launch, which faded, and July 2026, which held and kept growing.

020K40KMay 25LaunchAug 25Nov 25Feb 26May 26DevPass + blogSep 26
By product / people per month

Gateway carries the audience. DevPass arrived fast.

Gateway30,036 in Sep
May 25Sep 26
DevPass8,713 in Sep
May 25Sep 26
Docs2,865 in Sep
May 25Sep 26
Lounge2,199 in Sep
May 25Sep 26

Every panel shares one scale (top = 40K), so small panels are genuinely small.

People overlap across sites. Airside has no production pageviews in this project.

By referrer / people per quarter

Google compounds. Social does not.

Google40,071 in Q3 26
2025-Q22026-Q3
Direct / unavailable44,612 in Q3 26
2025-Q22026-Q3
Social & communities6,222 in Q3 26
2025-Q22026-Q3
Other search5,618 in Q3 26
2025-Q22026-Q3
Other sites4,152 in Q3 26
2025-Q22026-Q3
AI assistants849 in Q3 26
2025-Q22026-Q3
GitHub612 in Q3 26
2025-Q22026-Q3

Every panel shares one scale (top = 50K), so small panels are genuinely small.

Q2 2025 covers only May 30 to June 30. Direct includes app bookmarks and stripped referrers.

Gateway site content / people per month

Catalogue pages and tools grew. Comparison pages stalled.

Grouped by URL path on llmgateway.io, dashboard and auth pages removed.

Model pages9,276 in Sep
May 25Sep 26
Blog5,426 in Sep
May 25Sep 26
Data tools5,396 in Sep
May 25Sep 26
Provider pages2,441 in Sep
May 25Sep 26
Guides1,254 in Sep
May 25Sep 26
Changelog765 in Sep
May 25Sep 26
Compare & migration611 in Sep
May 25Sep 26

Every panel shares one scale (top = 10K), so small panels are genuinely small.

AI assistants / people per month

Small, but the fastest-rising source.

0300600May 25Sep 25Jan 26May 26Sep 26
AssistantJun 26Jul 26Aug 26Sep 26
ChatGPT99191249212
Gemini38100171180
Claude28315221
Perplexity19254335
Copilot / Grok2502
Tagged campaigns / utm_source, all period

Newsletters worked once each.

chatgpt.com2,237
tldrfounders2,092
convertkit1,366
tldrwebdev516
theneurondaily.com366
openalternative.co333
google284
blog236
substack207
perplexity118
tinyemail117

People with that tag on any pageview. chatgpt.com is added by ChatGPT itself, not a campaign we ran.

Self-reported source / since 2026-02-16

Half the answers say “Other”.

Other1,240
Twitter / X487
Reddit274
Email189
Product Hunt80
LinkedIn73
devntell29
Enterprise SSO5

2,377 answers at sign-up. The options leave out Google, AI assistants and newsletters, which the referrer data shows are the main sources.

Product milestones / people per quarter

Key creation grew fastest.

QuarterSign-up eventOnboarding doneAPI key createdEnterprise leads
2025-Q24315170
2025-Q33821021200
2025-Q45251541890
2026-Q16088534080
2026-Q29091,98088413
2026-Q31,2588673,10717

Monthly people summed per quarter. The sign-up event under-reports (see decision 05). Onboarding was redesigned in February 2026, which explains its jump. Paid-plan starts are excluded as commercial data.

Tracking timeline

Most funnel events are recent.

  • page_viewed_signup, page_viewed_login
  • user_signed_up, api_key_created
  • onboarding_viewed, referral_source_selected
  • cta_clicked, pricing_plan_clicked
  • nav_link_clicked
  • model_survey_* (Model Census)
  • conversion_rail_shown

Anything before these dates is missing measurement, not zero activity. Full-period funnels are not possible from PostHog alone.

Decision 01

Keep feeding the search engine.

Google-referred people grew every quarter, from 2,985 in Q3 2025 to 40,071 in Q3 2026. No other named channel compounds like this.

Prioritise catalogue-backed pages (models, providers, rankings, calculators) and refresh them when models ship. Pair with Search Console on equal windows before claiming ranking gains.

Decision 02

Build tools and pages, not one-off posts.

Model pages reached 9,276 people in September and data tools 5,396, both from near zero. Compare and migration pages have been flat at roughly 600–900 a month since February.

Rework the comparison and migration set around current search intent before adding more of them. Measure each by people and next step, not by count published.

Decision 03

Treat launches and newsletters as spikes.

The July 2025 launch month reached 9,441 people and fell to 3,148 two months later. Social and community referrals then stayed near 2,400 a quarter for a year. Newsletter tags (TLDR Founders, ConvertKit, TLDR WebDev) each brought one burst.

Plan each launch with a landing page that keeps earning search traffic afterwards, and send spike visitors to one clear first step.

Decision 04

Measure AI-assistant discovery properly.

AI-assistant referrals rose from 10 people in June 2025 to 450 in September 2026. Gemini went from 10 to 180 in six months. ChatGPT links also arrive tagged utm_source=chatgpt.com (2,237 people), which the referrer count misses.

Merge referrer and UTM signals into one AI-assistant channel and track it monthly next to Google.

Decision 05

Fix the sign-up signals first.

PostHog's user_signed_up covers about 16% of all-time admin sign-ups, and 52% of the "where did you hear about us" answers are "Other".

Fire the sign-up event on every auth path (OAuth, DevPass, Lounge) and add Google, AI assistant and newsletter options to the survey. Until then, no funnel rates.

Full-period data and measurement limits
  • People are PostHog person_id counts per month. A person active in several months is counted in each; months do not add up to a total audience.
  • Raw HogQL does not apply the project test-account filter. Staff traffic is included; flagged bots are shown and removed from marketing pageviews.
  • Marketing pageviews exclude /dashboard paths on every host. Pageviews from July 2026 onward are mostly logged-in product use, so people is the better trend line.
  • Referrer buckets use each pageview referrer, not first touch. ChatGPT also tags links with utm_source=chatgpt.com, so AI referrals by domain undercount.
  • user_signed_up reached 3,723 people in PostHog against 23,308 all-time admin sign-ups. The event misses most sign-up paths; use it for trend shape only.
  • Many events began mid-period (see tracking timeline). Absence before a start date is missing instrumentation, not zero activity.
  • Paid-plan starts, purchases, top-ups, refunds, balances and any monetary properties were not collected.
MonthPeopleSessionsPageviewsOutside /dashboardFlagged bots
May 253345113700
Jun 251,5221,9985,6874,17937
Jul 259,44110,78622,64519,30947
Aug 254,7186,08019,09714,80148
Sep 253,1484,47015,86711,99212
Oct 253,6585,11317,10313,73418
Nov 257,1409,21324,35419,44421
Dec 255,2267,04022,10417,63046
Jan 265,6457,57024,15119,870104
Feb 268,21310,90339,26529,95295
Mar 268,77511,82741,53130,431255
Apr 2612,30316,33061,03643,061436
May 2613,81820,47269,89748,286478
Jun 2616,65727,61598,53662,401414
Jul 2628,60460,348230,557124,120915
Aug 2632,84587,052340,806162,2621,005
Sep 2637,01182,400295,234150,411901

PostHog MCP · project LLMGateway (163518) · HogQL execute-sql · queried 2026-09-27 · analytics.json

B / The latest week

One week,
in detail.

September 20–26 against the week before, plus the Search Console and admin snapshots captured the same day. Most of the pageview dip was DevPass. Financial data is excluded.

Three different scopes. PostHog: September 20–26 versus September 13–19, 2026, UTC. Search Console: the supplied 7-day preset, aggregated weekly, with its native reporting lag. Admin: all-time Audience totals. Do not combine these into one funnel.

Production pageviews

61,649-22.5% vs previous week

Previous: 79,576
Production-host pageviews, not sessions

Tracked visitor IDs

11,343-3.1% vs previous week

Previous: 11,710
Tracked identifiers, not deduplicated people

CTA click events

802-35.1% vs previous week

Previous: 1,236
Project-wide named events, not a matched cohort

Recorded signup IDs

100-21.3% vs previous week

Previous: 127
Tracked identifiers, not deduplicated people

Diagnose the DevPass decline before adding more assets.

DevPass accounts for 80.8% of the observed pageview decrease. Its tracked visitor IDs fell from 2,829 to 2,202. Gateway’s IDs were almost flat. Check traffic mix, returning usage and instrumentation before attributing the change to marketing.

ProductPageviews
previous → current
Tracked IDs
previous → current
Pageview change
Gateway
llmgateway.io
36,282 → 33,1679,524 → 9,513-8.6%
DevPass
devpass.llmgateway.io
39,509 → 25,0182,829 → 2,202-36.7%
Docs
docs.llmgateway.io
2,282 → 2,107865 → 806-7.7%
Lounge
lounge.llmgateway.io
1,503 → 1,357374 → 335-9.7%

Identifiers overlap across products; do not add product counts to estimate people. Airside has no row in this whitelisted pageview sample, which is not proof of zero visitors.

Google Search Console / observed search

Brand demand, plus useful discovery queries.

4.02KClicks · rounded UI label
220KImpressions · rounded UI label
1.8%Reported average CTR
7Reported average position
Visible queryClicksImpressions
llm gateway4962,449
devpass232398
llmgateway151377
atria api61417
typesafe ai402,642
llm gateway devpass3052
atria asi24522
gateway llm1861

First visible query rows, not the full query population or a brand-share estimate. Single-window observations do not establish search improvement. Last update: 4.5 hours ago at capture.

PostHog / public content

Decision-led content earns attention.

Current UTC week, Gateway host only. These are observed pageviews and identifiers, not completed setup or attributable conversions.

Sampled pagePageviewsTracked IDs
/4,2002,692
/models1,9121,057
/pricing773621
/blog/best-ai-coding-plans653535
/blog/scx-model-discount-extended202158
/blog/scx-model-discount150123
/blog/minimax-h3-max-rapper-swap-video125105
/providers11670
/blog/opencode-go-pricing8778
/guides/claude-code7554

The coding-plan comparison article leads the sampled editorial pages. Test its route to a successful setup before producing more variants.

Referrers / descriptive, not attribution

Search is visible. Campaign coverage needs a better denominator.

Direct / unavailable34,590
Google14,880
Product family / internal8,493
Other external1,704
Other search1,178
Social / community403
AI assistants231
GitHub170

Pageview referrers, not acquisition channels. “Google” is not a verified organic-only segment. AI-assistant referrers show 231 pageviews from 61 IDs, not measured citations.

UTM source appears on 858 of 61,649 pageviews (1.4%); campaign on 100 (0.16%). This is field presence, not tracking completeness: organic/direct visits need no tags, and tags may not persist on later pages.
Engagement / events are not a funnel

Measure the path between these stages.

Observed stageEventsTracked IDs
page_viewed_code_landing841484
page_viewed_signup1,197964
user_signed_up100100
onboarding_viewed223166
api_key_created450367
onboarding_api_key_copied10071
Conversion rail: 1,155 shown events from 1,032 IDs; 11 click events from 11 IDs. These are unjoined counts, not a cohort click-through rate. Validate matching exposure/click identities before testing a clearer next step.

Key creation includes existing users and multiple keys. Do not divide these stage totals into a signup or activation conversion rate.

Admin Audience / all time

A larger audience than a single-week cohort.

23,308Total sign-ups
20,225Verified users
30,542Organizations

All-time platform counts. Verification is not first-use activation; organizations are not unique people. Do not compare these totals directly with a seven-day event cohort. The financial cards and the Paying row were excluded from collection.

Measurement boundaries and source receipts
  • distinct_id counts are tracked identifiers, not deduplicated people; bot and staff traffic was not independently audited.
  • Pageviews are filtered to the declared production host list. Selected named events use the project-wide event stream, so their populations are not identical.
  • Named stage counts are unjoined. Do not divide signup, CTA and key-creation counts into a conversion funnel.
  • Referrer buckets describe pageview referrers, not first-touch acquisition or last-non-direct session attribution.
  • Two adjacent full UTC weeks show observed movement, not causation, seasonality control or an experiment.
  • Airside produced no row in the whitelisted pageview sample. This does not establish zero visitors or correct instrumentation.
  • Google headline counts are preserved as displayed, with no invented extra precision. Its weekly bucket labels are not used as UTC interval boundaries.
  • Admin totals are a stock of accounts, not a weekly acquisition or retention cohort.
  • All changes are descriptive observations. No experiment, causal uplift or financial outcome is claimed.

Only non-financial aggregate search, pageview, engagement and account counts. No revenue, spend, balances, invoices, monetary properties, customer records or session recordings collected for this update. Published at the owner's request as a static page. Financial data, personal data and unpublished local source are excluded.

01 / The portfolio

One ecosystem.
Different buying decisions.

The shared infrastructure is useful. The marketing still has to make a separate promise to each audience.

01

LLM Gateway

Platform and engineering teams
One governed route to AI infrastructure

Next action
A first API request or an enterprise conversation

Routing, observability, spend controls, security and deployment choices

Local homepage refresh
02

DevPass

Developers using coding agents
A coding-focused plan and a familiar toolchain

Next action
Plan evaluation, signup and agent setup

Agent guides, model comparisons, pricing and usage views

Implemented in source
03

Lounge

People using AI directly
A named workspace for chat and creative studios

Next action
Try a studio, then choose a membership

Visible guest introductions, comparison pages and real interface assets

Implemented in source
04

Airside

Inference providers
A self-serve route into the catalog

Next action
Register a provider and complete verification

Provider onboarding, model verification, pricing review and provider guides

Implemented in source
02 / What exists

Count the work.
Do not call it growth.

These counts are source artifacts, not impressions, visitors, customers or campaign performance. Full file inventory below.

Blog posts98
Changelog entries109
Setup guides23
Migration guides5
Use-case articles4
DevPass comparisons6

245 editorial files at the snapshot. Comparison routes generated from code are additional and not included in this file count.

4Product tour recordings
1Enterprise demo with captions
PairedLight and dark product assets

Product screenshots are omitted from this non-financial report refresh. The presentation uses non-financial diagrams and the same approved evidence snapshot.

03 / The marketing work

A system with
sixteen connected parts.

Each workstream pairs the implementation with a transferable lesson and the metric needed to evaluate it.

01 / PositioningSource + local refresh

A portfolio, not four disconnected websites

  • Distinct audiences and product narratives for Gateway, DevPass, Lounge and Airside.
  • Shared product switching connects the portfolio across applications.
  • Enterprise-first homepage work leads with a production pilot and makes the other products discoverable.
The takeawayShare the distribution surface, not an indistinguishable pitch. Give each audience its own proof and next action.
Measure next: Qualified activation by product and entry page
02 / PositioningSource + local refresh

Names and visual identity became product infrastructure

  • The chat product was renamed Lounge with a coherent membership vocabulary and identity.
  • Brand assets, social images and paired light/dark screenshots make product identity reusable.
  • A richer product screenshot family and interactive homepage demo are local work at this snapshot.
The takeawayA new name must reach navigation, metadata, docs, pricing and imagery together. A metaphor cannot replace the plain product definition.
Measure next: Correct product recognition and destination selection in usability checks
03 / DiscoveryImplemented in source

An owned editorial engine

  • 98 blog files and 109 changelog files preserve launches, tutorials, comparisons, benchmarks and product updates.
  • Category pages, changelog tags and pagination make the archive navigable.
  • 23 setup guides, 5 migration guides and 4 use-case articles turn product capabilities into search and evaluation material.
The takeawayShipping is raw material. Convert a release into a problem-led article, a setup path and a clear product action.
Measure next: Matched content-entry to qualified first-success cohorts; start with the most-visited decision pages
04 / DiscoveryImplemented in source

Pages for people already choosing a solution

  • Gateway comparison pages address alternatives across routers, cloud platforms and coding tools.
  • Six DevPass comparison articles and Lounge comparisons serve distinct buying decisions.
  • Token and Copilot cost calculators, model categories, provider pages and country pages create useful discovery entry points.
The takeawayMatch one search intent to one useful page. Offer a decision aid, not a thin template with a changed keyword.
Measure next: Comparison or calculator use to setup or purchase intent
05 / DiscoverySource + observed search

Search foundations were audited across products

  • Per-product robots, sitemaps, canonical metadata and structured data are in source.
  • The September 5 audit records guest-visible Lounge content, account noindex fixes, fewer duplicate headings and truthful sitemap dates.
  • The historical audit reports 86 rendered-page checks. This update separately observes Search Console’s displayed 7-day search metrics; it does not claim SEO lift.
The takeawayVisible useful content and crawlable canonical pages come before search hacks. Treat an old audit as dated evidence.
Measure next: Index coverage, non-brand qualified search visits and field performance
06 / DiscoveryImplemented in source

Discovery for agents as well as humans

  • Public product context and pricing summaries ship as machine-readable files.
  • Docs expose markdown mirrors and full-context endpoints alongside human pages.
  • The prior audit removed stale catalog promises. PostHog now records AI-assistant referrer activity, which is not evidence of citation frequency or search-engine visibility gains.
The takeawayKeep the HTML page and machine-readable mirror consistent. Context files support consumption; they do not prove AI-search visibility.
Measure next: A repeatable query/citation baseline with dated responses
07 / ActivationImplemented in source

Documentation became a product-specific activation path

  • Docs are split into Gateway, DevPass, Lounge and Airside sections.
  • The snapshot contains 99 Gateway, 31 DevPass, 10 Lounge and 3 Airside MDX pages.
  • Integration guides and migration instructions reduce the gap between an article and a first successful action.
The takeawayAwareness without a working next step leaks demand. Documentation is part of the conversion path, not an afterthought.
Measure next: Guide-to-first-success rate and time to first success
08 / EvaluationImplemented assets

Product proof became visible and reusable

  • Paired product screenshots show dashboards and studios across light and dark themes.
  • Four product tour MP4s exist under docs recordings; an enterprise demo also ships with chapters and captions.
  • Benchmark articles link protocols and results, including the smart-routing comparison that explicitly reports limitations and losing baselines.
The takeawayShow the actual workflow. Keep benchmark method, full costs, failures and scope beside the result.
Measure next: Demo-assisted qualified activation; benchmark replication checks
09 / ConversionSource + observed events

Content has an explicit route into the product

  • Blog calls to action route relevant readers to DevPass with campaign parameters.
  • A contextual conversion rail records shown, dismissed and clicked events.
  • Observed PostHog events include 802 CTA clicks and 100 distinct signup IDs in the current UTC week. These counts are unjoined and are not a conversion rate.
The takeawayChoose the next action from reader intent. Instrument the action and connect it to downstream value before claiming uplift.
Measure next: CTA click to signup to qualified product activation
10 / ConversionImplemented in source

Sales assistance and trust surfaces

  • Enterprise pages expose security, deployment, procurement, pricing and support material.
  • A contact/scheduling surface and a chaptered demo support assisted evaluation.
  • Public licensing and legal explanations help buyers understand the offer instead of relying only on hero copy.
The takeawayFor enterprise buyers, evaluation friction is a marketing problem. Put procurement and deployment proof near the decision.
Measure next: Qualified inquiry to accepted pilot to paid conversion
11 / DistributionSource records; reach unmeasured

Launches and community distribution

  • A dated DevPass Product Hunt announcement links the launch destination.
  • Gateway launch, Lounge rebrand and Airside introduction have durable owned announcements.
  • Repository/community links and changelog publication provide distribution surfaces; social campaign activity outside the repo was not audited.
The takeawayAn announcement is an asset; distribution is a separate task. Record the destination, permission, date and attributable visits.
Measure next: Attributed qualified activation per launch destination
12 / DistributionImplemented in source

Provider promotions have a lifecycle

  • Provider launch promotions and extensions are documented in dated posts.
  • Shared promotion configuration carries start/end dates and a selector for the currently active offer.
  • The current tree includes an extended campaign ending October 9; this is configuration evidence, not proof of campaign performance or deployment.
The takeawayTreat an offer like software: one source of truth, exact dates, eligibility, automatic expiry and matching billing behavior.
Measure next: Incremental activated usage and contribution after discount
13 / RetentionImplemented in source

Newsletter and lifecycle automation

  • The newsletter form connects to a rate-limited subscription API and an email platform.
  • Worker follow-ups distinguish no-purchase, low-usage and no-repurchase behavior.
  • Marketing tips and credit alerts have separate categories, signed unsubscribe links and list-unsubscribe headers.
The takeawayBehavior should choose the message. Verify consent, suppression and unsubscribe at the send boundary; source code is not a delivery report.
Measure next: Incremental activation or retention with unsubscribe and complaint guardrails
14 / RetentionImplemented in source

A product-native referral loop

  • Public referral links and a referral handler connect acquisition to an organization.
  • A referral dashboard and sharing UI make the program discoverable inside the product.
  • Bonus computation is configurable and checks referral relationships; purchase gating is a separate contract to verify.
The takeawayA referral program needs usable sharing, correct reward rules and abuse controls. A route is not evidence that people referred others.
Measure next: Referred qualified accounts and retained value net of rewards
15 / TrustSource + local claims edits

Claim freshness is now an engineering concern

  • A shared marketing module centralizes headline claims across surfaces.
  • Support knowledge derives public catalog counts and includes billing guidance.
  • Repository rules explicitly reject copied catalog lists in marketing and docs; local claim edits remain separate from verified current facts.
The takeawayCentralization alone does not make a claim true. Define eligible records, scope, freshness and evidence, then reuse the same source.
Measure next: Percentage of public claims with an owner, evidence and review date
16 / MeasurementObserved non-financial aggregates

Live observations now complement the implementation

  • Search Console shows search demand; PostHog provides dated pageview and engagement counts; the admin Audience panel supplies all-time account totals.
  • PostHog compares September 20–26 with September 13–19 in UTC. Recorded pageviews, CTA events and signup IDs declined; this is not a controlled experiment.
  • Only non-financial aggregate metrics are included. Delivery reports, matched activation cohorts and causal attribution remain outside this evidence.
The takeawayAssets shipped, events implemented and business outcomes are different levels of evidence. Never turn one into the other.
Measure next: Matched, privacy-safe first-visit to signup to first-success cohorts
04 / A short history

From an API launch
to a product family.

Milestones are dated by content and source history. They establish repository records, not launch performance or production deployment.

Gateway introduction

Owned product narrative begins in the blog.

Version one

A public changelog creates a durable release record.

Consumer entry point

The chat product is introduced alongside the API.

DevPass launch distribution

A Product Hunt announcement connects to an external launch destination.

Referral loop

Referral and cache-control updates are recorded together.

A named portfolio

Lounge gets an identity; provider launch offers become time-bound campaigns.

Provider acquisition + search foundations

Airside introduction and a multi-product SEO audit.

Proof, product docs, enterprise focus

Benchmark evidence and product-specific docs; the homepage refresh is local work.

05 / What to repeat

The lessons worth
turning into a skill.

Not a universal list of tactics. A way to connect real product evidence to useful work and honest learning.

01

Separate the audiences

One portfolio can share navigation without forcing every buyer through the same pitch.

02

Turn releases into useful paths

Announcement, proof, guide and next action should connect.

03

Make claims maintainable

Use current eligible records, dated evidence and scheduled expiry.

04

Show the work, including the losses

Real demos and reproducible comparisons beat broad superiority claims.

05

Finish the conversion loop

Click events and campaign parameters are a start, not an attribution system.

06

Measure before scaling

Run a bounded experiment and report denominator, uncertainty and guardrails.

06 / What this audit cannot prove

Keep the unknowns
in the picture.

Absence from this checkout does not mean an activity never happened. It means the claim needs another authorized source.

Causality and activation

Search, engagement and audience aggregates are now observed. Matched acquisition-to-first-success cohorts, retention and causal lift remain unmeasured. Financial analysis is deliberately excluded.

Distribution completeness

Paid ads, outbound, social schedules, influencer work, press and event activity may exist elsewhere; no evidence was provided here.

Launch exports

Prior session memory mentions four rendered launch videos. Their project/export files were not found in this checkout; the four docs tour MP4s and enterprise video were found.

Deployment

A file or merged commit does not establish the current production deployment. Homepage work was concurrent and uncommitted at the baseline.

Comparability

The September SEO audit explicitly says its local LCP runs selected different elements and do not establish a speedup.

Freshness

Historical pricing and availability in announcements must not be repeated as current promises. Shared constants also require validation.

Different windows and populations

Search Console uses its own 7-day preset with weekly buckets and rounded headlines. PostHog compares two full UTC weeks; admin totals are all time. These figures must not be combined into a funnel.

07 / Recommended sequence

Measure one path.
Improve one leak.

These are recommendations, not campaigns already run. Choose a product and an outcome before adding more assets.

08 / Take it with you

The presentation.

18 slides built from the same repository evidence and PostHog history as this report.

09 / The audit trail

Inspect the evidence.

Baseline commit 01da451e5c75. Counts are generated from the file inventory; relevant local source is frozen separately. Historical titles may contain obsolete claims and are included only as records.

Editorial inventory · 245 files
DateTypeHistorical title / source file
2026-06-02comparisonsDevPass vs Alibaba Qwen Coding Plan
apps/code/src/content/comparisons/alibaba-qwen-coding.md
2026-06-14comparisonsDevPass vs Cursor
apps/code/src/content/comparisons/cursor.md
2026-06-02comparisonsDevPass vs FirePass by Fireworks
apps/code/src/content/comparisons/firepass.md
2026-08-24comparisonsDevPass vs OpenCode Go
apps/code/src/content/comparisons/opencode-go.md
2026-06-02comparisonsDevPass vs OpenCode Zen
apps/code/src/content/comparisons/opencode-zen.md
2026-06-02comparisonsDevPass vs z.ai GLM Coding Plan
apps/code/src/content/comparisons/z-ai-glm-coding-plan.md
2026-01-25blogBeyond Proxies: Why Enterprises Need a Unified AI Gateway
apps/ui/src/content/blog/2025-01-25-unified-ai-gateway.md
2026-01-25blogWhat is an LLM Gateway?
apps/ui/src/content/blog/2025-01-25-what-is-an-llm-gateway.md
2025-04-12blogIntroducing LLM Gateway
apps/ui/src/content/blog/2025-04-12-introducing-llm-gateway.md
2025-05-01blogHow to Self-Host LLM Gateway
apps/ui/src/content/blog/2025-05-01-self-host-llm-gateway.md
2025-05-10blogCustom OpenAI-Compatible Providers Are Now Supported
apps/ui/src/content/blog/2025-05-10-custom-providers.md
2025-09-08blogHow to Configure Claude Code to Use Any Model via LLM Gateway
apps/ui/src/content/blog/2025-09-08-how-configure-claude-code-with-llmgateway.md
2025-10-14blogIntroducing LLM Gateway Chat — Test 280+ Models in One Interface
apps/ui/src/content/blog/2025-10-14-introducing-llmgateway-playground.md
2025-11-28blogBlack Friday at LLM Gateway – 50% Off Credits
apps/ui/src/content/blog/2025-11-28-black-friday-credits-promo.md
2025-12-01blogCyber Monday at LLM Gateway – 50% Off Credits
apps/ui/src/content/blog/2025-12-01-cyber-monday-credits-promo.md
2025-12-18blogQuarter 4 2025: Major Platform Updates
apps/ui/src/content/blog/2025-12-18-q4-2025.md
2025-12-27blogDeploying Next.js on GCP
apps/ui/src/content/blog/2025-12-27-nextjs-on-gcp.md
2026-02-03blogWhy Your AI App Needs a Gateway Layer
apps/ui/src/content/blog/2026-02-03-why-your-ai-app-needs-a-gateway.md
2026-02-05blogThe Hidden Cost of LLM Vendor Lock-in
apps/ui/src/content/blog/2026-02-05-hidden-cost-of-llm-vendor-lock-in.md
2026-02-08blogGetting Started with LLM Gateway in 5 Minutes
apps/ui/src/content/blog/2026-02-08-getting-started-in-5-minutes.md
2026-02-10blog20% Off All Alibaba Cloud Qwen Models on LLM Gateway
apps/ui/src/content/blog/2026-02-10-alibaba-cloud-qwen-models-20-percent-off.md
2026-02-11blogOpenAI vs Anthropic vs Google: Real Cost Comparison 2026
apps/ui/src/content/blog/2026-02-11-openai-vs-anthropic-vs-google-cost-comparison.md
2026-02-14blogHow We Cut Our LLM Costs 60% With Request Routing
apps/ui/src/content/blog/2026-02-14-cut-llm-costs-with-request-routing.md
2026-03-23blogQ1 2026: Video Gen, Image Studio & Enterprise Features
apps/ui/src/content/blog/2026-03-23-q1-2026.md
2026-03-28blogUp to 30% Off DeepSeek Models on LLM Gateway
apps/ui/src/content/blog/2026-03-28-deepseek-discounts-on-llm-gateway.md
2026-03-28blogTop 10 Cheapest Providers for DeepSeek V3.2 in 2026
apps/ui/src/content/blog/2026-03-28-top-10-cheapest-providers-deepseek-v3-2.md
2026-04-09blog8 Best AI Gateways in 2026 (Compared)
apps/ui/src/content/blog/2026-04-09-best-ai-gateways.md
2026-04-11blogHow to Choose the Right LLM for Your Use Case in 2026
apps/ui/src/content/blog/2026-04-11-how-to-choose-the-right-llm.md
2026-04-11blogHow We Handle LLM Provider Failover at Scale
apps/ui/src/content/blog/2026-04-11-how-we-handle-llm-provider-failover.md
2026-04-11blogLLM Gateway vs OpenRouter: An Honest Comparison
apps/ui/src/content/blog/2026-04-11-llm-gateway-vs-openrouter.md
2026-04-11blogLLM Guardrails Explained: Prompt Injection, PII Detection & Content Moderation
apps/ui/src/content/blog/2026-04-11-llm-guardrails-explained.md
2026-04-18blogLLM Gateway Is Now a Built-in Provider in OpenCode
apps/ui/src/content/blog/2026-04-18-opencode-built-in-provider.md
2026-04-23blogLLM Gateway vs LiteLLM: An Honest Comparison
apps/ui/src/content/blog/2026-04-23-llm-gateway-vs-litellm.md
2026-04-24blogHow to Estimate LLM Token Costs Before You Ship
apps/ui/src/content/blog/2026-04-24-how-to-estimate-llm-token-costs.md
2026-04-25blogPrompt Caching Explained: How to Cut LLM Costs by 30–99%
apps/ui/src/content/blog/2026-04-25-prompt-caching-explained.md
2026-04-26blogLLM Gateway vs Direct API: When the Provider SDK Stops Scaling
apps/ui/src/content/blog/2026-04-26-llm-gateway-vs-direct-api.md
2026-05-15blogEmbeddings on LLM Gateway: One API for Vectors and Chat
apps/ui/src/content/blog/2026-05-15-openai-compatible-embeddings.md
2026-05-17blogByteDance Seedance Lands on LLM Gateway: Three Video Models, One API
apps/ui/src/content/blog/2026-05-17-seedance-video-generation.md
2026-05-26blogLLM Gateway vs Portkey: An Honest Comparison
apps/ui/src/content/blog/2026-05-26-llm-gateway-vs-portkey.md
2026-05-26blogWhat Is LLM Orchestration? Patterns, Tools & When You Need One
apps/ui/src/content/blog/2026-05-26-what-is-llm-orchestration.md
2026-06-07blogStripe for AI: Embeddable Payments for Your App
apps/ui/src/content/blog/2026-06-07-embeddable-payments-sdk.md
2026-06-10blogSpeech Generation Is Live: ElevenLabs, OpenAI & Gemini TTS Through One API
apps/ui/src/content/blog/2026-06-10-speech-generation-and-audio-studio.md
2026-06-11blogLLM Gateway Is Now SOC 2 Type II Compliant
apps/ui/src/content/blog/2026-06-11-soc2-type-ii.md
2026-06-18blogAPI Key Rotation: How We Secure Your API Keys
apps/ui/src/content/blog/2026-06-18-api-key-rotation.md
2026-06-18blogBuilding a Slack Q&A Bot with LLM Gateway and Chat SDK
apps/ui/src/content/blog/2026-06-18-slack-qa-bot-with-llm-gateway.md
2026-06-20blogHow to Deploy LLM Gateway on Cloud Platforms
apps/ui/src/content/blog/2026-06-20-deploy-llmgateway-on-cloud-platforms.md
2026-06-22blog11 Best AI Coding Plans in 2026 (Compared)
apps/ui/src/content/blog/2026-06-22-best-ai-coding-plans.md
2026-06-22blog10 Best ChatGPT Alternatives in 2026
apps/ui/src/content/blog/2026-06-22-best-chatgpt-alternatives.md
2026-06-23blogAI Gateway Pricing: Fees & Markups Compared (2026)
apps/ui/src/content/blog/2026-06-23-ai-gateway-fees-compared.md
2026-06-23blogClaude Opus 4.8 Pricing and the Hidden Tokenizer Tax
apps/ui/src/content/blog/2026-06-23-claude-opus-tokenizer-cost.md
2026-06-26blogEnterprise LLM Analytics: See Where Every Dollar Goes
apps/ui/src/content/blog/2026-06-26-enterprise-llm-analytics.md
2026-06-30blogQ2 2026: Speech, Embeddings & Coding Plans
apps/ui/src/content/blog/2026-06-30-q2-2026-roundup.md
2026-07-01blogDevPass Code: A Terminal Coding Agent for LLM Gateway
apps/ui/src/content/blog/2026-07-01-devpass-code.md
2026-07-05blogProjects: A Knowledge Base for Your AI Chats
apps/ui/src/content/blog/2026-07-05-chat-projects-knowledge-base.md
2026-07-08blog8 Best LiteLLM Alternatives in 2026 (Compared)
apps/ui/src/content/blog/2026-07-08-litellm-alternatives.md
2026-07-12blog8 Best GitHub Copilot Alternatives in 2026 (Compared)
apps/ui/src/content/blog/2026-07-12-github-copilot-alternatives.md
2026-07-12blogMicrosoft Copilot Enterprise Pricing in 2026, Explained
apps/ui/src/content/blog/2026-07-12-microsoft-copilot-enterprise-pricing.md
2026-07-19blog9 Best Open-Source LLMs in 2026 (Compared)
apps/ui/src/content/blog/2026-07-19-best-open-source-llms.md
2026-07-19blogHow to Use Kimi K3 with Claude Code, Cursor, and Cline
apps/ui/src/content/blog/2026-07-19-kimi-k3-claude-code.md
2026-07-19blogKimi K3 vs Claude Opus 4.8: Benchmarks, Price, Verdict
apps/ui/src/content/blog/2026-07-19-kimi-k3-vs-claude-opus.md
2026-07-19blogKimi K3 and China's Open-Weight Model Wave
apps/ui/src/content/blog/2026-07-19-kimi-k3.md
2026-07-22blogOpenRouter vs Vercel vs LLMGateway Performance
apps/ui/src/content/blog/2026-07-22-openrouter-vs-vercel-vs-llmgateway-performance.md
2026-07-23blog6 Best Open-Source OpenRouter Alternatives in 2026
apps/ui/src/content/blog/2026-07-23-open-source-openrouter-alternatives.md
2026-07-23blogOpenRouter Alternatives for Enterprise Teams (2026)
apps/ui/src/content/blog/2026-07-23-openrouter-alternatives-for-enterprise.md
2026-07-23blog10 Best OpenRouter Alternatives in 2026 (Compared)
apps/ui/src/content/blog/2026-07-23-openrouter-alternatives.md
2026-07-26blogHow to Generate Audio with a Text-to-Speech API
apps/ui/src/content/blog/2026-07-26-generate-audio-api.md
2026-07-26blogHow to Generate Images with One API (2026 Guide)
apps/ui/src/content/blog/2026-07-26-generate-images-api.md
2026-07-26blogHow to Generate AI Videos with an API
apps/ui/src/content/blog/2026-07-26-generate-videos-api.md
2026-07-26blogHow to Track LLM Usage and Spend with the API
apps/ui/src/content/blog/2026-07-26-track-llm-usage-spend-api.md
2026-08-04blogGDPR-Compliant LLM Routing, Enforced at the Gateway
apps/ui/src/content/blog/2026-08-04-gdpr-compliant-llm-routing.md
2026-08-04blog8 Best Helicone Alternatives in 2026 (Compared)
apps/ui/src/content/blog/2026-08-04-helicone-alternatives.md
2026-08-04blogKimi K3 API Guide: Pricing, Reasoning, and Setup
apps/ui/src/content/blog/2026-08-04-kimi-k3-api.md
2026-08-04blogKimi K3 Open Weights Are Out: What You Can Actually Do
apps/ui/src/content/blog/2026-08-04-kimi-k3-open-weights.md
2026-08-04blogThe LLM Compliance Checklist for Production Teams
apps/ui/src/content/blog/2026-08-04-llm-compliance-checklist.md
2026-08-04blogLLM Data Retention: What to Store, and for How Long
apps/ui/src/content/blog/2026-08-04-llm-data-retention.md
2026-08-04blog8 Best Portkey Alternatives in 2026 (Compared)
apps/ui/src/content/blog/2026-08-04-portkey-alternatives.md
2026-08-08blogRanked #1 on an Independent AI Gateway Benchmark
apps/ui/src/content/blog/2026-08-08-ai-gateway-benchmark.md
2026-08-12blogA Vercel AI Gateway Alternative Without the Rewrite
apps/ui/src/content/blog/2026-08-12-vercel-ai-gateway-alternative.md
2026-08-15blogIs Kimi Down? How to Check Kimi K3 API Status
apps/ui/src/content/blog/2026-08-15-kimi-api-status.md
2026-08-15blogWhat Is an AI Coding Plan? Limits and Costs Explained
apps/ui/src/content/blog/2026-08-15-what-is-an-ai-coding-plan.md
2026-08-16blogStripe's OpenRouter Acquisition: What Changes for You
apps/ui/src/content/blog/2026-08-16-stripe-openrouter-acquisition.md
2026-08-18blogAzure AI Foundry vs GitHub Copilot vs One Gateway
apps/ui/src/content/blog/2026-08-18-azure-ai-foundry-vs-github-copilot.md
2026-08-21blogTanStack AI + LLM Gateway: One Adapter, 200+ Models
apps/ui/src/content/blog/2026-08-21-tanstack-ai-llm-gateway.md
2026-08-24blog9 Best Cursor Alternatives in 2026 (Compared)
apps/ui/src/content/blog/2026-08-24-cursor-alternatives.md
2026-08-24blogHow to Use GLM-5.2 with Claude Code, Cursor, and Cline
apps/ui/src/content/blog/2026-08-24-glm-5-2-claude-code.md
2026-08-24blogOpenCode Go's New Pricing, Explained
apps/ui/src/content/blog/2026-08-24-opencode-go-pricing.md
2026-09-05blogVideo Generation With the Vercel AI SDK 7 and LLM Gateway
apps/ui/src/content/blog/2026-09-05-ai-sdk-7-video-generation.md
2026-09-05blogCache-Aware LLM Routing Learns Your Workload
apps/ui/src/content/blog/2026-09-05-cache-aware-llm-routing.md
2026-09-05blogEnterprise CLI: SSO Login, Organization Skills, 15 Coding Agents
apps/ui/src/content/blog/2026-09-05-enterprise-cli-browser-login-organization-skills.md
2026-09-05blogIntroducing Airside: List Your Models on LLM Gateway
apps/ui/src/content/blog/2026-09-05-introducing-airside.md
2026-09-05blogCheck LLM Usage and Costs From Your Coding Agent
apps/ui/src/content/blog/2026-09-05-mcp-llm-usage-analytics-guide.md
2026-09-09blogSCX Model Discount: 20% Off Through October 9
apps/ui/src/content/blog/2026-09-09-scx-model-discount.md
2026-09-13blogHow to Get Your LLM API Listed on an AI Gateway
apps/ui/src/content/blog/2026-09-13-get-listed-on-ai-gateway.md
2026-09-13blogHow LLM Routing Picks a Provider: A Carrier's Guide
apps/ui/src/content/blog/2026-09-13-llm-routing-carriers-guide.md
2026-09-13blogHow to Make Money With LLM Inference Through Airside
apps/ui/src/content/blog/2026-09-13-make-money-llm-inference.md
2026-09-14blogCheap vs Flagship AI Models (financial amounts omitted)
apps/ui/src/content/blog/2026-09-14-ai-coding-model-benchmark.md
2026-09-20blogAn LLM Decision API That Returns Values, Not Text
apps/ui/src/content/blog/2026-09-20-llm-decision-api.md
2026-09-20blogThe Perplexity Sonar API Retirement: What Changes
apps/ui/src/content/blog/2026-09-20-perplexity-sonar-api-retirement.md
2026-09-22blogRecreate the Rapper-Swap Video Trend with MiniMax H3 Max
apps/ui/src/content/blog/2026-09-22-minimax-h3-max-rapper-swap-video.md
2026-09-22blogSCX Model Discount Extended Through October 9
apps/ui/src/content/blog/2026-09-22-scx-model-discount-extended.md
2026-09-23blogCompliance Alerts for Blocked AI Models
apps/ui/src/content/blog/2026-09-23-compliance-alerts.md
2026-09-25blogAutomatic Model Selection by Request Difficulty
apps/ui/src/content/blog/2026-09-25-automatic-model-selection.md
2026-09-25blogLLM Cost Optimization at k a Month (financial amounts omitted)
apps/ui/src/content/blog/2026-09-25-llm-routing-savings-100k.md
2026-09-26blogSmart Routing Benchmark: Cost and Quality
apps/ui/src/content/blog/2026-09-26-smart-routing-benchmark.md
2025-08-07changelogGPT-5 Model Family Now Available
apps/ui/src/content/changelog/2025-01-26-gpt-5-models-support.md
2025-08-11changelogAPI Key Usage Limits & Credit Controls
apps/ui/src/content/changelog/2025-01-27-api-key-usage-limits.md
2025-08-16changelogNew Models Directory & Free Llama 3.1 70B via CloudRift
apps/ui/src/content/changelog/2025-01-28-models-page-cloudrift-free-llama.md
2025-05-01changelogLLM Gateway v1.0 Launch
apps/ui/src/content/changelog/2025-05-01-gateway-v1-launch.md
2025-05-11changelogDeepseek Discount + Mistral Joins the Lineup
apps/ui/src/content/changelog/2025-06-11-deepseek-discount-and-mistral-models.md
2025-06-14changelogSelf-Hosting Just Got Easier
apps/ui/src/content/changelog/2025-06-14-self-hosting-docs-upgrade.md
2025-06-15changelogPro Subscription Launch
apps/ui/src/content/changelog/2025-06-15-pro-subscription-launch.md
2025-06-22changelogDashboard UI Improvements & Project Context
apps/ui/src/content/changelog/2025-06-22-dashboard-ui-improvements.md
2025-06-26changelogNew Providers: Groq and xAI Support + Always-Visible Credits
apps/ui/src/content/changelog/2025-06-26-groq-xai-providers-and-sidebar-credits.md
2025-07-13changelogNew Providers: Cloudrift, Moonshot AI and Novita AI Support
apps/ui/src/content/changelog/2025-07-13-moonshot-novita-cloudrift-providers.md
2025-07-28changelogNext.js migration
apps/ui/src/content/changelog/2025-07-28-nextjs-migration.md
2025-08-05changelogNew GPT-OSS Models: 120B and 20B via Groq
apps/ui/src/content/changelog/2025-08-05-gpt-oss-models-groq.md
2025-08-06changelogClaude 4.1 Models: Opus and Sonnet Now Available
apps/ui/src/content/changelog/2025-08-06-claude-4-1-models.md
2025-08-07changelogAI SDK Provider v2.0 Released
apps/ui/src/content/changelog/2025-08-07-ai-sdk-provider-v2.md
2025-08-24changelogDeepSeek v3.1 Model Support
apps/ui/src/content/changelog/2025-08-24-deepseek-v3-1-support.md
2025-08-26changelogIntroducing Our First Image Generation Model: Gemini 2.5 Flash Image Preview
apps/ui/src/content/changelog/2025-08-26-gemini-2-5-flash-image-preview.md
2025-09-05changelogQwen3 Max Model Now Available
apps/ui/src/content/changelog/2025-09-05-qwen3-max-support.md
2025-09-08changelogClaude Code Configuration Now Supported
apps/ui/src/content/changelog/2025-09-08-claude-code-configuration-support.md
2025-09-11changelogNew Alibaba Qwen Models: Qwen3 Next, Max, Plus, Flash, Vision
apps/ui/src/content/changelog/2025-09-11-alibaba-qwen-models.md
2025-09-20changelogGrok 4 Fast Models: Flagship and Fast Variants Now Available
apps/ui/src/content/changelog/2025-09-20-grok-4-fast-models.md
2025-09-29changelogClaude Sonnet 4.5 Model Support
apps/ui/src/content/changelog/2025-09-29-claude-sonnet-4-5-support.md
2025-10-18changelogCanopyWave Partnership: 90% Off DeepSeek v3.1
apps/ui/src/content/changelog/2025-10-18-canopywave-partnership-deepseek-90-off.md
2025-10-24changelogTeam Members: Roles, Seats, and Access Controls
apps/ui/src/content/changelog/2025-10-24-team-members-roles-and-seats.md
2025-10-26changelogAWS Bedrock, Google Vertex AI and Microsoft Azure
apps/ui/src/content/changelog/2025-10-26-aws-bedrock-vertex-azure-providers.md
2025-11-05changelogZ.ai 10% Off & Google 20% Off All Models
apps/ui/src/content/changelog/2025-11-05-zai-google-discounts.md
2025-11-07changelogKimi K2 Thinking Model Support
apps/ui/src/content/changelog/2025-11-07-kimi-k2-thinking-model.md
2025-11-08changelogCanopyWave: 3 New Models with 75% Off
apps/ui/src/content/changelog/2025-11-08-canopywave-new-models.md
2025-11-11changelogCanopyWave: 75% Off Kimi K2 Thinking
apps/ui/src/content/changelog/2025-11-11-canopywave-kimi-k2-thinking-75-off.md
2025-11-17changelogSherlock: Two New Stealth Alpha Models
apps/ui/src/content/changelog/2025-11-17-sherlock-stealth-models.md
2025-11-18changelogGemini 3 Pro Preview: 20% Off Launch Discount
apps/ui/src/content/changelog/2025-11-18-gemini-3-pro-preview-support.md
2025-12-14changelogCerebras: Ultra-Fast Inference with 6 New Models
apps/ui/src/content/changelog/2025-12-14-cerebras-models-support.md
2026-01-02changelogAlibaba Cloud Qwen Image Models: Advanced Image Generation and Editing
apps/ui/src/content/changelog/2026-01-02-qwen-image-max-support.md
2026-01-21changelogPro Features Now Free for Everyone
apps/ui/src/content/changelog/2026-01-21-pro-features-now-free.md
2026-01-26changelogEnterprise Audit Logs
apps/ui/src/content/changelog/2026-01-26-enterprise-audit-logs.md
2026-01-26changelogEnterprise Guardrails
apps/ui/src/content/changelog/2026-01-26-enterprise-guardrails.md
2026-01-29changelogDev Plans, Native Web Search, and MiniMax Provider
apps/ui/src/content/changelog/2026-01-29-dev-plans-web-search-minimax.md
2026-01-30changelogAI Agent skills, Agents, Templates & CLI
apps/ui/src/content/changelog/2026-01-30-agents-templates-cli.md
2026-01-30changelogUnified Reasoning Configuration
apps/ui/src/content/changelog/2026-01-30-reasoning-max-tokens.md
2026-02-12changelogAutomatic Retry & Fallback with Full Routing Transparency
apps/ui/src/content/changelog/2026-02-12-retry-fallback.md
2026-02-27changelogImage Studio, Image Edits API & More
apps/ui/src/content/changelog/2026-02-27-image-studio-and-more.md
2026-03-06changelogGPT-5.4 and GPT-5.4 Pro Now Available
apps/ui/src/content/changelog/2026-03-06-gpt-5-4-models.md
2026-03-23changelogVideo Generation, Sessions & More
apps/ui/src/content/changelog/2026-03-23-video-gen-sessions-and-more.md
2026-04-02changelogMulti-Region Routing, Content Filters & More
apps/ui/src/content/changelog/2026-04-02-multi-region-content-filters.md
2026-04-12changelogSessions Rebranded to Agents
apps/ui/src/content/changelog/2026-04-12-sessions-rebranded-to-agents.md
2026-05-07changelogDevPass is live on Product Hunt
apps/ui/src/content/changelog/2026-05-07-devpass-product-hunt-launch.md
2026-05-15changelogOpenAI-Compatible Embeddings
apps/ui/src/content/changelog/2026-05-15-openai-compatible-embeddings.md
2026-05-17changelogSeedance Video Models, Pinned Chats, Sharing Across Orgs & More
apps/ui/src/content/changelog/2026-05-17-seedance-chat-product-roundup.md
2026-05-26changelogDocument Reading (PDFs & more)
apps/ui/src/content/changelog/2026-05-26-document-reading-support.md
2026-06-01changelogClaude Opus 4.8 + a Wave of New Models
apps/ui/src/content/changelog/2026-06-01-claude-opus-4-8-and-new-models.md
2026-06-01changelogReferrals, Cache Controls & Product Updates
apps/ui/src/content/changelog/2026-06-01-referrals-cache-controls-product-updates.md
2026-06-01changelogSmarter Routing: Sticky Sessions, Bedrock Regions & More
apps/ui/src/content/changelog/2026-06-01-smarter-routing-sticky-sessions.md
2026-06-09changelogAPI Key Expiration (TTL)
apps/ui/src/content/changelog/2026-06-09-api-key-expiration-ttl.md
2026-06-10changelogChat Plans, Service Tiers, SDK Sandbox & More
apps/ui/src/content/changelog/2026-06-10-chat-plans-service-tiers-product-roundup.md
2026-06-10changelogClaude Fable 5, Reve Image Gen & More New Models
apps/ui/src/content/changelog/2026-06-10-claude-fable-5-reve-and-new-models.md
2026-06-10changelogSpeech Generation + Audio Studio
apps/ui/src/content/changelog/2026-06-10-speech-generation-and-audio-studio.md
2026-06-13changelogClaude Fable 5 Access Suspended
apps/ui/src/content/changelog/2026-06-13-claude-fable-5-access-suspended.md
2026-06-14changelogRouting Strategies: Cheapest, Fastest & Defaults
apps/ui/src/content/changelog/2026-06-14-smart-routing-strategies.md
2026-06-16changelogProvider Compliance Policies
apps/ui/src/content/changelog/2026-06-16-provider-compliance-policies.md
2026-06-18changelogCustom Model Catalog for Custom Providers
apps/ui/src/content/changelog/2026-06-18-custom-model-catalog.md
2026-06-21changelogUsage Analytics by Model, Key and Member
apps/ui/src/content/changelog/2026-06-21-org-analytics.md
2026-06-26changelogOrganization-Wide Analytics
apps/ui/src/content/changelog/2026-06-26-organization-analytics.md
2026-07-03changelogPer-Member Budgets, Developer Role & PDF Invoices
apps/ui/src/content/changelog/2026-07-03-team-budgets-product-roundup.md
2026-07-09changelogGPT-5.6 Sol, Terra & Luna Now Available
apps/ui/src/content/changelog/2026-07-09-gpt-5-6-models.md
2026-07-12changelogProvider Headquarters Compliance Filter
apps/ui/src/content/changelog/2026-07-12-provider-headquarters-filter.md
2026-07-16changelogDevPass Reset Passes
apps/ui/src/content/changelog/2026-07-16-devpass-reset-passes.md
2026-07-22changelogUpgrade Rollover, New Providers & Gemini TTS
apps/ui/src/content/changelog/2026-07-22-upgrade-rollover-new-providers.md
2026-07-23changelogLaunch Any Coding Agent from the CLI
apps/ui/src/content/changelog/2026-07-23-cli-launch-coding-agents.md
2026-07-24changelogChat Is Now Lounge
apps/ui/src/content/changelog/2026-07-24-lounge-rebrand.md
2026-07-26changelogRealtime Voice API
apps/ui/src/content/changelog/2026-07-26-realtime-voice-api.md
2026-07-27changelogRunware Launch: 30% Off Open-Source Models
apps/ui/src/content/changelog/2026-07-27-runware-launch-discount.md
2026-07-30changelogGPT-5.6 Terra & Luna Price Cuts
apps/ui/src/content/changelog/2026-07-30-gpt-5-6-terra-luna-price-cut.md
2026-07-31changelogOrg Models Directory
apps/ui/src/content/changelog/2026-07-31-org-models-directory.md
2026-08-03changelogModerations Pricing From August 7
apps/ui/src/content/changelog/2026-08-03-moderations-flat-request-pricing.md
2026-08-05changelogDevPass Pay-As-You-Go Overflow
apps/ui/src/content/changelog/2026-08-05-devpass-payg-overflow.md
2026-08-07changelogDynamic Routes
apps/ui/src/content/changelog/2026-08-07-dynamic-routes.md
2026-08-10changelogPay For Credits With Crypto
apps/ui/src/content/changelog/2026-08-10-crypto-payments.md
2026-08-12changelogDrop-In For The Vercel AI Gateway
apps/ui/src/content/changelog/2026-08-12-ai-sdk-gateway-protocol.md
2026-08-17changelogTrust Tiers: Limits That Grow With Your Account
apps/ui/src/content/changelog/2026-08-17-org-trust-tiers-limits.md
2026-08-21changelogClient-Managed Prompt Caching
apps/ui/src/content/changelog/2026-08-21-client-managed-cache-control.md
2026-08-22changelogConcurrent Request Limits and Overload Protection
apps/ui/src/content/changelog/2026-08-22-concurrent-request-limits.md
2026-08-22changelogProvider Key Descriptions, Grok 4.6 & More
apps/ui/src/content/changelog/2026-08-22-provider-key-descriptions-new-models.md
2026-08-27changelogHash-Only API Key Storage
apps/ui/src/content/changelog/2026-08-27-hash-only-key-storage.md
2026-08-28changelogClaude on Foundry, Model Status & More
apps/ui/src/content/changelog/2026-08-28-claude-foundry-model-status.md
2026-08-28changelogRemove Your DevPass Payment Card
apps/ui/src/content/changelog/2026-08-28-devpass-card-removal.md
2026-08-28changelogNo AI Training for DevPass
apps/ui/src/content/changelog/2026-08-28-devpass-no-ai-training.md
2026-08-28changelogOrganization Teams and Directory Sync
apps/ui/src/content/changelog/2026-08-28-organization-teams.md
2026-09-01changelogAirside: Self-Serve Provider Listings
apps/ui/src/content/changelog/2026-09-01-airside-provider-console.md
2026-09-01changelogClaude Fable 5.1
apps/ui/src/content/changelog/2026-09-01-claude-fable-5-1.md
2026-09-04changelogZero Data Retention Controls
apps/ui/src/content/changelog/2026-09-04-zero-data-retention.md
2026-09-05changelogAI SDK 7 Provider with Video Generation
apps/ui/src/content/changelog/2026-09-05-ai-sdk-7-provider-video-generation.md
2026-09-05changelogAirside: Model Verification and Crew Invites
apps/ui/src/content/changelog/2026-09-05-airside-model-verification.md
2026-09-05changelogAdaptive Cache-Aware Provider Selection
apps/ui/src/content/changelog/2026-09-05-cache-aware-routing.md
2026-09-05changelogCLI Browser Login and Organization Skills
apps/ui/src/content/changelog/2026-09-05-cli-browser-login-organization-skills.md
2026-09-05changelogCompare Usage Across Periods
apps/ui/src/content/changelog/2026-09-05-dashboard-usage-comparison.md
2026-09-05changelogGemini 3.8 Flash, Muse Spark 1.3 & More Models
apps/ui/src/content/changelog/2026-09-05-gemini-3-8-flash-muse-spark-new-models.md
2026-09-05changelogUsage Analytics in the MCP Server
apps/ui/src/content/changelog/2026-09-05-mcp-usage-analytics.md
2026-09-05changelogPer-Unit Price Filters, Seedream References & More
apps/ui/src/content/changelog/2026-09-05-per-unit-price-filters-product-roundup.md
2026-09-06changelogRealtime Transcription Sessions
apps/ui/src/content/changelog/2026-09-06-realtime-transcription-sessions.md
2026-09-08changelogGPT Image 2.5 Is Here
apps/ui/src/content/changelog/2026-09-08-gpt-image-2-5.md
2026-09-18changelogNative VS Code Integration
apps/ui/src/content/changelog/2026-09-18-vscode-extension.md
2026-09-19changelogPerplexity Sonar Moves to the Agent API
apps/ui/src/content/changelog/2026-09-19-perplexity-agent-api.md
2026-09-20changelogPerplexity Sonar Changes September 25
apps/ui/src/content/changelog/2026-09-20-perplexity-sonar-changes.md
2026-09-20changelogSystem One: Typed Decisions
apps/ui/src/content/changelog/2026-09-20-system-one-typed-decisions.md
2026-09-22changelogClaude Opus 5.5
apps/ui/src/content/changelog/2026-09-22-claude-opus-5-5.md
2026-09-22changelogGPT-6 Sol & Luna at Half the Price
apps/ui/src/content/changelog/2026-09-22-gpt-6-sol-luna.md
2026-09-23changelogCompliance Alerts
apps/ui/src/content/changelog/2026-09-23-compliance-alerts.md
2026-09-25changelogBranded Payments SDK Receipts
apps/ui/src/content/changelog/2026-09-25-branded-end-user-receipts.md
2026-09-25changelogSmart Routing
apps/ui/src/content/changelog/2026-09-25-smart-routing.md
2026-09-26changelogAzure Priority Processing
apps/ui/src/content/changelog/2026-09-26-azure-priority-processing.md
2026-09-23guidesAnvil Integration
apps/ui/src/content/guides/anvil.md
2026-09-23guidesAutohand Code Integration
apps/ui/src/content/guides/autohand.md
2026-09-07guidesClaude Code Integration
apps/ui/src/content/guides/claude-code.md
2026-09-23guidesCline Integration
apps/ui/src/content/guides/cline.md
2026-09-07guidesCodex CLI Integration
apps/ui/src/content/guides/codex-cli.md
2026-09-23guidesContinue CLI Integration
apps/ui/src/content/guides/continue.md
2026-09-23guidesCrush Integration
apps/ui/src/content/guides/crush.md
2026-09-23guidesCursor Integration
apps/ui/src/content/guides/cursor.md
2026-09-07guidesDevPass Code Integration
apps/ui/src/content/guides/devpass-code.md
2026-09-23guidesEmpryo Integration
apps/ui/src/content/guides/empryo.md
2026-09-23guidesGitHub Copilot App Integration
apps/ui/src/content/guides/github-copilot.md
2026-09-07guidesHermes Agent Integration
apps/ui/src/content/guides/hermes-agent.md
2026-09-23guidesKilo Code Integration
apps/ui/src/content/guides/kilo-code.md
2026-09-23guidesKimi Code Integration
apps/ui/src/content/guides/kimi-code.md
2026-07-03guidesMCP Server Integration
apps/ui/src/content/guides/mcp.md
2026-09-23guidesMiMo Code Integration
apps/ui/src/content/guides/mimocode.md
2026-09-23guidesn8n Integration
apps/ui/src/content/guides/n8n.md
2026-09-22guidesOh My Pi Integration
apps/ui/src/content/guides/oh-my-pi.md
2026-09-23guidesOpenClaw Integration
apps/ui/src/content/guides/openclaw.md
2026-09-23guidesOpenCode Desktop Integration
apps/ui/src/content/guides/opencode-desktop.md
2026-09-07guidesOpenCode Integration
apps/ui/src/content/guides/opencode.md
2026-09-07guidesPi Coding Agent Integration
apps/ui/src/content/guides/pi.md
2026-09-07guidesReasonix Integration
apps/ui/src/content/guides/reasonix.md
2026-07-12migrationsMigrate from GitHub Copilot
apps/ui/src/content/migrations/github-copilot.md
2026-01-20migrationsMigrate from LiteLLM
apps/ui/src/content/migrations/litellm.md
2026-01-20migrationsMigrate from OpenRouter
apps/ui/src/content/migrations/openrouter.md
2026-05-26migrationsMigrate from Portkey
apps/ui/src/content/migrations/portkey.md
2026-01-20migrationsMigrate from Vercel AI Gateway
apps/ui/src/content/migrations/vercel-ai-gateway.md
2026-06-02use-casesAI customer support
apps/ui/src/content/use-cases/ai-customer-support.md
2026-06-02use-casesCoding agents & AI assistants
apps/ui/src/content/use-cases/coding-agents.md
2026-06-02use-casesAI cost optimization & FinOps
apps/ui/src/content/use-cases/cost-optimization.md
2026-06-02use-casesRAG & document Q&A
apps/ui/src/content/use-cases/rag-document-qa.md
Documentation inventory · 143 pages

gateway / 99

devpass / 31

lounge / 10

airside / 3

Source registry · 63 references
S01 · Local work at snapshotapps/ui/src/app/page.tsxLocal work, not published
S02 · Local work at snapshotapps/ui/src/components/home/home-hero.tsxLocal work, not published
S36 · Repository sourceapps/ui/src/content/blog/2026-09-26-smart-routing-benchmark.mdInspect pinned source
S41 · Local work at snapshotapps/ui/src/components/home/tracked-link.tsxLocal work, not published
S45 · Repository sourceapps/ui/src/content/changelog/2026-05-07-devpass-product-hunt-launch.mdInspect pinned source
S46 · Repository sourceapps/ui/src/content/changelog/2025-05-01-gateway-v1-launch.mdInspect pinned source
S49 · Local work at snapshotpackages/shared/src/marketing.tsLocal work, not published
S50 · Repository sourceapps/ui/src/content/changelog/2026-07-27-runware-launch-discount.mdInspect pinned source
S51 · Repository sourceapps/ui/src/content/blog/2026-09-22-scx-model-discount-extended.mdInspect pinned source
S53 · Local work at snapshotapps/ui/src/components/landing/newsletter.tsxLocal work, not published
S62 · Local work at snapshotapps/ui/src/components/home/control-plane.tsxLocal work, not published
S63 · Repository sourceapps/playground/src/components/seo/lounge-landing-sections.tsxInspect pinned source

Non-financial live observations: analytics.json. Repository evidence: evidence.json. Frozen source excerpts and local implementations: a frozen local snapshot (not published). The report generator and verification scripts live beside this file.