ChatGPT¶

Development line: project:chatgpt · thread chatgpt-product-development
Last event: 2025-07-30 · 5 dated since 2023-08-04 · Researched: 2026-09-04 · confidence: medium

What it is¶

ChatGPT is an AI assistant for drafting text, analyzing files and images, researching current topics, and interactive learning.

• Text: writes, explains, translates, and reasons.
• Files and images: reads files and images.
• Web search: searches the web with citations.
• Images: creates and edits images.
• Workflows: runs Voice, Study Mode, data analysis, Projects, and Scheduled Tasks where enabled.

Models, tools, and message caps change by plan, region, workspace, and settings. Use it as a working interface to several AI tools, not as an authority for unverified facts.

Development line¶

• 2023-08-04 — OpenAI published a ChatGPT feature update. On 2023-08-04, OpenAI published the linked first-party ChatGPT feature announcement. The source marks it as feature development without giving the announcement text, so capability, audience, and rollout status remain unstated.
• 2023-09-25 — OpenAI introduced multimodal interaction in ChatGPT. On 2023-09-25, OpenAI announced that ChatGPT would accept visual and audio input and provide spoken responses. This expanded the product beyond typed chat into multimodal interaction through an active rollout.
• 2024-04-02 — OpenAI documented DALL·E image editing in ChatGPT. On 2024-04-02, OpenAI help documentation described an image-editing workflow using DALL·E. Together with the announcement, it established image editing as a product capability without specifying account eligibility.
• 2024-12-21 — OpenAI announced phone access for ChatGPT. On 2024-12-21, OpenAI announced phone access for ChatGPT. This added a separate access channel, though geographic reach and calling terms were not detailed.
• 2025-07-30 — OpenAI introduced Study Mode in ChatGPT. On 2025-07-30, OpenAI announced Study Mode for ChatGPT. This guided learning interactions rather than only supplying direct answers, without detailing plan or regional availability.

What changed¶

ChatGPT expanded from basic chat conveniences into multimodal workflows and integrated tools.

• 2023-04-27: change unknown; the external page could not be read and no first-party equivalent was found.
• 2023-08-04: an OpenAI release note dated 2023-08-03 added prompt examples, suggested replies, model persistence for Plus users, multiple-file analysis in Code Interpreter, and keyboard shortcuts.
• 2023-09-25: voice conversation and image input began rolling out to Plus and Enterprise users; voice was mobile-only at launch and image input was planned for all platforms.
• 2023-10-30: change unknown; no URL was retained.
• 2024-04-02: image editing appeared under DALL·E; the current path is ChatGPT Images, with natural-language edits and a selection tool.
• 2024-12-17: change unknown; no URL was retained.
• 2024-12-21: experimental access through 1-800-CHATGPT extended the product beyond the app; current availability is region-specific.
• 2025-07-30: Study Mode added guided, Socratic, step-by-step learning instead of only giving an answer.
• 2026-09-04: first-party documentation describes an assistant with web search, deep research, image input and editing, data analysis, Voice, Projects, and Scheduled Tasks; availability remains account-dependent.

Historical rollout statements do not establish present access. The durable change is from a chat interface to configurable multimodal workflows.

How to use this¶

Choose interaction features by release date: use visual, audio, and spoken interaction from 2023-09-25, DALL·E image editing from 2024-04-02, phone access from 2024-12-21, and Study Mode from 2025-07-30.

1. Open chatgpt.com or the official mobile app, select New chat, state your task, and send it. — https://help.openai.com/en/articles/12677804-what-is-chatgpt-faq
2. Give the goal, key context, desired output, and constraints; review the answer, then refine the prompt. — https://help.openai.com/en/articles/10032626-how-do-i-prompt-chatgpt-effectively
3. For time-sensitive questions, select the web-search icon or use /Search, then open the cited sources. — https://help.openai.com/en/articles/12677804-what-is-chatgpt-faq
4. For images, select Images or upload a file; for a local edit, select the area and describe the change. — https://help.openai.com/en/articles/11084440-chatgpt-images
5. For a spoken session, select Voice, grant microphone access, choose an available option, and start speaking. — https://help.openai.com/en/articles/20001274
6. For learning, choose Tools → Study and learn, state your goal and level, then work through its questions. — https://openai.com/index/chatgpt-study-mode/
7. For a one-off private context, start a Temporary Chat and choose personalized or non-personalized behavior before sending content. — https://help.openai.com/en/articles/8914046-temporary-chat-faq

Best practices¶

• Make prompts clear, specific, and contextual; iterate after inspecting the first result. — https://help.openai.com/en/articles/10032626-how-do-i-prompt-chatgpt-effectively
• Verify date-sensitive, time-sensitive, or location-sensitive claims, and supply exact dates, time zones, or locations when relevant. — https://help.openai.com/en/articles/20001274
• Use Search for live information, then inspect cited pages rather than treating the generated answer as the source. — https://help.openai.com/en/articles/12677804-what-is-chatgpt-faq
• Before sharing sensitive material, review Data Controls or use a Temporary Chat; temporary chats can still be retained for up to 30 days for safety. — https://help.openai.com/en/articles/7730893-chatgpt-shared-links-faq
• For image edits, name the target area and inspect the output: selection highlights are not always exact and edits can extend outside them. — https://help.openai.com/en/articles/11084440-chatgpt-images

Superseded by this¶

• 2023-09-25: “Voice is a Plus/Enterprise mobile beta enabled from Settings → New Features” is obsolete. Voice now starts from the Voice icon on supported web and mobile experiences; availability and limits vary.
• 2024-04-02: “Edit images in ChatGPT through the official DALL·E GPT” is obsolete. The official DALL·E GPT is retired; use ChatGPT Images.
• 2024-12-21: “Phone or WhatsApp access is one universal ChatGPT channel” is obsolete. Calls require a U.S. or Canadian number in supported countries, while current WhatsApp availability is EEA-specific and may roll out gradually.

Still unknown¶

• The 2023-04-27 Reddit link could not be read, and no first-party source established its exact product change.
• The 2023-10-30 and 2024-12-17 entries have no retained URL, leaving their specific changes unattributed.
• The linked X posts were not directly readable; their exact wording remains unknown even where first-party documentation corroborates related functionality.
• Current access, models, and limits were not tested on a specific account or locale and can differ by plan, region, workspace, app version, and settings.
• Phone access and image editing fit ChatGPT as access and capability areas, not as separate subjects.

Sources¶

Agent brief¶

• Subject: project:chatgpt, thread chatgpt-product-development, 5 dated events 2023-08-04 → 2025-07-30.
• Practical note: Choose interaction features by release date: use visual, audio, and spoken interaction from 2023-09-25, DALL·E image editing from 2024-04-02, phone access from 2024-12-21, and Study Mode from 2025-07-30.
• Confidence: medium. Dated supersedes above are the authority for what is obsolete.