ChatGPT Ads Context Hints and Multi-Image Ads: An APAC Operating Guide
OpenAI has published guidance for ad-group context hints and opened multi-image ads in beta. Hints describe relevance rather than enforce targeting, while 4 to 6 images share one destination URL.
NEWS · AI SEARCH LABOpenAI opened multi-image ChatGPT Ads in beta on October 26, 2026 and published context-hint guidance on October 28. Context hints add natural-language details about the offer, audience need and useful situation at ad-group level. They inform relevance but do not act as targeting rules. Multi-image ads use four to six square images, all pointing to one URL with the current Learn more CTA.
What changed on October 26 to 28?
Ads Manager now offers single-image and multi-image formats for standard campaigns using Reach, Clicks or Conversions objectives. Advertisers can also add context hints to describe when an ad may be relevant. The features solve different problems: one broadens the visual story, while the other adds semantic detail.
More images do not automatically raise relevance. More hints cannot force delivery. The creative, hint and landing page need to describe the same proposition. APAC teams should begin with one customer situation per ad group, then verify that every market destination supports it.
Write hints around What, Who and When
OpenAI recommends three kinds of detail. What covers features, services, pricing, availability and service areas. Who covers needs, preferences and circumstances. When covers the decision or situation in which the offer is useful. Write one clear idea per hint in natural language instead of a disconnected keyword list.
The facts must remain current. Update hints when prices, promotions, availability or service areas change. Exact word matching is not required, so repeated synonyms add less value than a precise condition a buyer can verify.
Review a hint now. A secure annual plan for mid-sized companies operating Seoul offices is more useful than a loose enterprise, secure, Seoul and discount list, provided the landing page supports the claim.
Hints are relevance inputs, not targeting commands
OpenAI states that hints cannot enforce geography, schedules or exclusions, and do not guarantee an appearance. Use the separate controls available in Ads Manager for delivery settings. If two use cases need different messages or destinations, split them into separate ad groups rather than stretching one hint.

This distinction matters across Korea, Japan, Taiwan, Hong Kong, India and Australia. A regional service description may be shared, but price, legal entity, language, delivery and support can differ. Describe the local proposition; do not write instructions that pretend to control the system.
The official multi-image specification
A multi-image ad requires four to six square images at a 1:1 ratio. Every image is clickable, but all share one destination URL and one set of UTM parameters. The current CTA is Learn more. Headline and description requirements match the single-image format.
This is not a carousel with a different product URL behind every card. The images should provide distinct views of one proposition. OpenAI recommends clear variants, accurate and useful claims, a specific landing page rather than a generic homepage, and simple relevant imagery.
Do not duplicate one picture merely to meet the count. Use complementary views of the situation, outcome, configuration or comparison criterion, without adding benefits that the destination cannot substantiate.
Why the landing page and crawlers matter
Suggested ad drafts can use existing website metadata to prefill an image, title and description. OpenAI says this does not generate new copy or imagery with AI. The destination must not block OAI-AdsBot or OAI-SearchBot.
Crawler access is not a promise of ad delivery or organic citation. The page still needs visible facts about price, availability, geography, limitations and update date. Open the mobile destination and make sure the claim in every image and hint can be checked without returning to the homepage.
Separate paid relevance from GEO and AEO
Context hints are inputs to paid-ad relevance. They are not metadata for organic ChatGPT citations, Google AI Overviews, AI Mode or NAVER AI Tab. Measure paid impressions and clicks separately from organic citations, shopping appearances and conversions.
The shared requirement is a specific page that answers the user’s question. Paid teams check consistency between hint, creative and destination. Search teams check public evidence, crawl access, entity clarity and freshness. A common page can serve both, but the scorecards should remain separate.
Enterprise, healthcare and commerce checks
Large enterprises should map each ad group to its hint, creative, destination, market and approver. Keep a regional intent framework while validating currency, inventory, returns, legal entity and support locally. Split the group when the proposition changes.
Healthcare organisations must not use hints to work around sensitive-category or patient-targeting restrictions. Platform policy and local healthcare-advertising review remain necessary. This article is not legal or medical advice.
Commerce teams should update stock, price, shipping coverage and returns in the hint and page together. Because every image points to one URL, that page must substantiate the complete visual story.
AI Search Lab view
The update shows why conversational advertising needs an explainable customer situation rather than a large keyword pile. Keep hints concise and verifiable, manage creative and destinations through the same change process, and compare paid relevance with organic AI visibility without merging the metrics.
LeadGenLab’s GEO and AEO services review ChatGPT ads, organic answers and shopping results as separate systems. Global teams can scope Korea-first or wider APAC operations through the project contact.
Official sources
- https://help.openai.com/en/articles/20001521-writing-context-hints-for-chatgpt-ads
- https://help.openai.com/en/articles/20001212-creating-ads-for-chatgpt-ads
Checked October 28, 2026 against OpenAI’s official help pages.
분석을 실제 실행으로 연결하려면
AI Search Lab은 공식 출처와 편집 기준에 따라 변화와 실행 기준을 설명합니다. 진단, 기술·콘텐츠 개선, 산업별 GEO·AEO 지원이 필요하다면 운영사 LeadGenLab의 관련 서비스와 상담 페이지에서 다음 단계를 확인하세요.