September Pro Library summary
The headline totals are aggregate lower bounds from the full observed corpus re-extracted on September 6, 2026. This publication record does not expose the underlying Pro rows.
Pro Library corpus statistics
ChatGPT Ads statistics
The strongest figures from the full Pro Library corpus, published here as a citation-ready view of ChatGPT advertising activity.
September 2026 data. Re-extracted and verified on September 6, 2026, covering ads observed from July through September 2026. Headline totals use the September corpus; niche rankings and creative benchmarks use the dated public July 2026 browse snapshot. Every figure is an independent lower-bound observation, not OpenAI's complete inventory.
11,103
Advertisers observed
43,410
Unique creatives
415,289
Captured placements
970
Niches covered
136,798
Associated prompts
45,356
Context hints
ChatGPT Ad Library has observed at least 11,103 advertisers, 43,410 unique creatives, and 415,289 sponsored placements across 970 niches. The full corpus connects those ads to 136,798 associated prompts and 45,356 inferred context hints.
Evidence record
These totals are published from a versioned lower-bound record, so the scope and date travel with the figures.
| Scope | full observed Pro Library corpus |
|---|---|
| Access | Published aggregate summary; row-level Pro evidence is not included |
| Observed through | September 6, 2026 |
| Extracted | September 6, 2026 |
| Revalidated | September 6, 2026 |
| Evidence ID | pro-library-corpus-2026-09-06 |
| Source version | serving-db-counts@2026-09-06-reextract |
| Public snapshot | July 2026 public browse snapshot (public-library-snapshot-2026-07-16) |
Two evidence scopes, kept separate
Public browse snapshot
Niche rankings and creative distributions use the separate July 2026 public browse snapshot, observed through July 16, 2026 and extracted on August 18, 2026.
How to read the numbers
| Metric | What it means |
|---|---|
| Unique creative | A normalized advertiser, title, copy, image, and destination combination. |
| Placement | One captured sponsored result for a probe prompt. A creative can produce multiple placements. |
| Trigger prompt | The prompt associated with an observed placement in the corpus. |
| Context hint | A reverse-engineered description of the conversational contexts an ad appears to target; not an advertiser's private account setting. |
Observed creative benchmarks
These describe distinct captured creatives in the public browse snapshot. They are not official OpenAI limits or performance recommendations.
28
Median title characters, n=16,639
63
Median description characters, n=16,610
99.5%
Creatives with an image, n=16,639
Largest niches in the public snapshot
These rankings use the public browse snapshot dated July 2026; the full corpus continues to grow separately.
Evidence change log
| Date | Record | Change |
|---|---|---|
| September 6, 2026 | pro-library-corpus-2026-09-06 | Full-corpus aggregate summary re-extracted and revalidated. |
| August 18, 2026 | public-library-snapshot-2026-07-16 | Public browse snapshot extracted for row-level rankings and creative distributions. |
Methodology and citation note
The library records ads returned while probing ChatGPT with realistic consumer and commercial prompts. It does not have access to OpenAI's internal campaign inventory, spend, budgets, targeting settings, or complete delivery logs. Counts are therefore lower-bound observations, not market totals.
FAQ
ChatGPT Ads statistics, answered
Direct answers with the limits attached.
How many ChatGPT ads have been captured?
The full observed corpus contains at least 43,410 unique ad creatives across 415,289 captured placements. These are independent observations, not OpenAI's complete campaign inventory.
How many advertisers has ChatGPT Ad Library observed?
ChatGPT Ad Library has observed at least 11,103 advertisers. The count describes this independently captured corpus and should not be interpreted as the total number of advertisers in OpenAI Ads Manager or as proof that every advertiser remains active.
What is a captured placement?
A placement is one observed sponsored result returned for a probe prompt. The same creative can appear for multiple prompts or on multiple dates, so placement count is larger than unique creative count.
Where do these ChatGPT Ads statistics come from?
They come from the documented collector-prompt methodology: recording sponsored placements returned for a controlled prompt set, then normalizing advertisers, creatives, niches, and associated prompts. The methodology page documents collection and coverage limits.
Can I cite these numbers?
Yes. Cite ChatGPT Ad Library, link to this page, state that the figures are an independently captured lower-bound sample, and include the access date. The headline corpus summary was revalidated on September 6, 2026; public browse rankings use the separate July 2026 snapshot.