You can now toggle GPT-5’s thinking time for faster or smarter answers – here’s how to do it (www.techradar.com)

OpenAI has added a thinking-time toggle to ChatGPT-5 for Plus and Pro users, letting you choose how long the model spends “thinking” before replying. On the web, Plus subscribers can switch between Standard (default for speed/intelligence balance) and Extended (more time for deeper answers). Pro users get two extra settings: Light (fastest, still using reasoning) and Heavy (maximum reasoning depth). To use it, select ChatGPT-5 → choose the Thinking model in the model selector, then pick the blue Thinking drop‑down in the message window. The feature is web-only for now; mobile apps will get it in coming weeks. Free users cannot access the Thinking modes—free access to GPT-5 remains limited to standard responses. This change matters because it gives users explicit control over the latency/quality trade-off of a large reasoning model. Light and Standard prioritize responsiveness for quick workflows and iteration, while Extended and Heavy allow more computational time for complex, multi-step tasks that benefit from deeper chain-of-thought processing. Practically, it affects prompt engineering, interactive apps, and research comparisons where time budget and answer fidelity matter. It may also influence cost and backend resource planning for OpenAI as variable inference times change throughput. Overall, the toggle improves usability by letting users choose faster or smarter answers depending on the task.