OpenAI widens GPT‑5 rollout

OpenAI is making GPT-5 the standard ChatGPT experience across web, mobile, and desktop while keeping workplace data connections under tighter admin control. (help.openai.com) OpenAI’s model release notes say GPT-5 is “slowly rolling out” worldwide to ChatGPT Plus, Pro, Team, and Free users, and that it is becoming the default model for logged-in users. The same note says ChatGPT Enterprise and Edu will get GPT-5 “soon,” rather than in the first wave. (help.openai.com) In OpenAI’s business release notes, the company says connectors remain on by default for ChatGPT Business, but off by default for Enterprise plans. In the Enterprise and Edu release notes, OpenAI says updated Box, Notion, Linear, and Dropbox apps are disabled by default, and admins must enable them in workspace settings. (help.openai.com, help.openai.com) Those connectors are now called “apps” in OpenAI’s help documentation, a naming change the company says took effect on December 17, 2025. OpenAI says the feature lets ChatGPT search, reference, or act inside outside services such as cloud storage and workplace software without leaving the chat. (help.openai.com) The split rollout shows two tracks inside ChatGPT. Consumer and small-team users are being pushed toward one default model, while larger organizations are still being asked to approve which external systems ChatGPT can touch, which groups can see them, and which actions are allowed. (help.openai.com, help.openai.com) OpenAI has been tightening those controls over the past two weeks. On April 9, 2026, Enterprise and Edu workspaces got a setting to hide Security Assertion Markup Language and System for Cross-domain Identity Management groups from sharing menus, which OpenAI said can reduce accidental oversharing in projects and custom generative pre-trained transformers. (help.openai.com) At the same time, OpenAI is swapping in faster fallback models for heavy use. On April 9, 2026, the company said GPT-5.3 Instant Mini replaced GPT-5 Instant Mini as the fallback model in Enterprise and Edu after users hit rate limits for GPT-5.3 Instant, and that the newer mini model will not appear in the model picker. (help.openai.com) OpenAI has framed GPT-5.3 Instant as the company’s most-used chat model, with a March 3, 2026 update aimed at tone, relevance, and conversational flow rather than a new benchmark headline. In the application programming interface changelog, OpenAI also said the GPT-5 family now includes smaller variants such as GPT-5 mini and GPT-5 nano for faster or higher-volume workloads. (openai.com, developers.openai.com) That combination — one default model for most chats, smaller fallback models when usage spikes, and admin-gated access to internal systems — points to OpenAI’s current playbook for business customers. The company is broadening GPT-5’s reach, but it is still making workspace owners decide when ChatGPT can read from Google Drive, Box, Notion, Dropbox, or other company tools. (help.openai.com, help.openai.com, help.openai.com)