OpenAI adds speed control to GPT-5 Thinking for Plus/Pro/Business

OpenAI has added a setting that lets Plus, Pro and Business users control how quickly GPT-5 with Thinking responds. The option appears in the message composer on the web and lets users trade off response speed for depth.

GPT-5 Thinking Pacing Modes

Subscribers can now select a pacing mode for GPT-5 Thinking from the message composer on the web. The chosen mode remains active across future chats on the web until users manually change it.

• Standard: The new default for all plans, offering a balance between speed and intelligence.
• Extended: The former default for Plus users; the model takes more time to produce deeper responses.
• Light: Exclusive to Pro users; delivers the snappiest replies for quick tasks.
• Heavy: Exclusive to Pro users; intended for the deepest reasoning with slower, more comprehensive outputs.

OpenAI Statement on GPT-5 Thinking

“We’ve heard your feedback that GPT-5 Thinking can sometimes take longer than you’d like. Now Plus, Pro, and Business users can set the pace to match the moment,” wrote the AI company on its X account.

Select GPT-5 with Thinking in ChatGPT on the web to toggle thinking time in the message composer.

GPT-5 Codex for Developers

OpenAI also unveiled GPT-5 Codex, a model designed for software development. According to the company, the model can write, debug and review code, operate interactively with humans or work independently on long projects, and handle simpler tasks with quick replies. OpenAI says GPT-5 Codex was trained to review code carefully and surface critical errors; tests on popular open-source projects showed its suggestions were more accurate and useful, helping developers focus on the most important issues.

Implications of GPT-5 Thinking for Users

The new control gives users a direct way to prioritise speed or depth depending on the task — quick factual lookups and short edits can run in Light/Standard, while complex reasoning or long-form analysis can use Extended or Heavy. For developers and teams, the combination of pacing control and a specialised code model (GPT-5 Codex) aims to offer more predictable resource and workflow choices on web-based ChatGPT.