ChatGPT’s iPhone app gained a document scanner on Oct. 1, 2026: users can photograph several paper pages in a row, and the app merges them into a single PDF that is ready to attach to a chat. OpenAI’s release notes describe the starting point as opening the camera inside ChatGPT, tapping the three-dot menu and choosing Scan. The entry is labeled for iOS and says nothing about Android, the web or any particular subscription plan.
The scanner arrived in the same stretch as a virtual try-on button for shopping, a new audio-upload tool and the GPT-6 rollout, which makes October’s entries unusually dense.
From camera to PDF inside a chat
The feature is documented in OpenAI’s ChatGPT release notes, the running log of changes to the product. The entry for Oct. 1 is brief, and the practical effect is to remove a step: instead of photographing pages, stitching them into a document in another app and then uploading the result, a user can capture the pages in ChatGPT’s own camera and get one file back. The release notes do not give a page limit, a file-size figure or a statement about which plans can use the scanner, so those details remain unconfirmed. Because the entry is one of dozens in a chronological log, a reader scanning the page for the feature has to look for the Oct. 1 date rather than finding it under a dedicated help article.
Uploads in general are governed by OpenAI’s file uploads FAQ. It sets a hard limit of 512MB per file and caps text and document files at 2 million tokens each. Users can upload up to 80 files every three hours, free users get three uploads per day, and OpenAI says those limits may be lowered during peak hours. A scanned PDF presumably falls under the same rules, though neither page says so outright.
One caveat sits in the same FAQ. ChatGPT Enterprise supports Visual Retrieval for PDFs, which lets it work with images inside them, while all other plans use text-based retrieval only: the system extracts digital text and discards images, for PDFs and other document files alike. The release notes do not say how a PDF built from camera photographs is read on those plans, so how well a scanned page of print or handwriting comes through is an open question that users will have to test for themselves.
iPhone owners already had a scanner of their own. Apple’s iPhone user guide lists scanning text and documents under the Notes app. What ChatGPT’s version changes is where the pages land: directly in the conversation, alongside whatever question the user wants to ask about them.
Try-on, audio uploads and GPT-6 in the same week
Also on Oct. 1, OpenAI added a “Try on” button to clothing and accessory listings in ChatGPT’s shopping results, on mobile and web. A user takes or uploads a selfie, and ChatGPT Images generates the preview; the reference photo is saved for later try-ons and can be changed or deleted under Settings, then Personalization, then Reference photos. OpenAI’s shopping help page cautions that results may not match the product or the person exactly and do not guarantee fit.
On Oct. 6 came audio uploads for paid plans. Users can attach an audio file to create transcripts, summarize recordings and ask questions about the content, such as turning a meeting, interview or lecture into structured notes. The release notes carry their own warning: “Transcripts may contain errors, and performance may vary across languages.”
The biggest item was GPT-6 with what OpenAI calls Intelligent UI, which began Oct. 7 for Plus, Pro, Business and Enterprise users, with Free and Go users following on Oct. 8. The paid plans use GPT-6 Sol and the free tiers use GPT-6 Luna. Intelligent UI lets ChatGPT mix text, visuals and interactive elements, such as side-by-side comparisons, calculators, bill splitters and games. The Pro reasoning option runs on GPT-6 Astra and does not support it. OpenAI adds that GPT-6 can begin answering while it keeps thinking or using tools, then add findings without another prompt, and that the models behind Work and Codex are not changing with this release.
September’s quieter changes, from Privacy Center to Pro 500
The scanner follows a run of September entries. On Sept. 21, OpenAI began rolling out a Privacy Center to signed-in Free, Go, Plus and Pro users. Interactive flashcards reached all plans on Sept. 22, and on Sept. 25 the web version gained a way to review recent security activity for an OpenAI account. On Sept. 29 the company introduced Pro 500, a $500-a-month plan with the highest included usage of its Pro plans.
No OpenAI spokesperson is quoted in the release notes, which are written as plain product entries, so the open questions are the practical ones: whether the scanner reaches Android, and which plans can use it beyond the iOS label the Oct. 1 entry gives.
This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.
More from Morning Overview
- Regulators cleared the first U.S. small modular reactor, 4 months early
- Lake Powell sank to a record 3,517 feet, nearing the level that stops Glen Canyon Dam’s turbines
- Seven of Earth’s nine planetary boundaries are breached, and all seven are worsening
- The NSA says three phone features should be off whenever you aren’t using them
