Prime Highlight
- OpenAI has enhanced ChatGPT with improved search reliability, cutting down hallucinations and delivering cleaner, more accurate results.
- The company also introduced speed controls for GPT-5 Thinking, allowing users to choose between four pacing modes for faster or more detailed responses.
Key Facts
- Pro, Plus, and Business subscribers can now select Standard, Extended, Light, or Heavy modes, with the chosen mode persisting across sessions.
- OpenAI launched GPT-5 Codex, a new coding model designed to debug, review, and test code with greater accuracy than previous versions.
Background
OpenAI has rolled out new updates to ChatGPT aimed at improving search reliability and giving users greater control over GPT-5 Thinking’s response speed. The company announced the changes on September 18 in a post on X.
The latest search improvements target three areas: factuality, shopping, and formatting. OpenAI said ChatGPT now delivers fewer hallucinations, provides better accuracy, and keeps shopping results more focused when product intent is detected. Answers are also displayed in cleaner formats, making them easier to read without losing detail.
Alongside search upgrades, OpenAI introduced speed control for GPT-5 Thinking. In addition, the Pro, Plus, and Business subscribers are now given the option of four pacing modes within the web composer: Standard (balanced output), Extended (more detailed responses), Light (swiftest replies), and Heavy (reasoning at its most detailed). The mode selected will remain active between sessions until the users manually switch it off.
The company claimed that this feature was created after some users mentioned that GPT-5 thinking is not as fast as it should be in certain situations. The update is flexible, and people can regulate speed according to their personal requirements to get a quick response or detailed reasoning.
OpenAI also introduced GPT-5 Codex, a developer-friendly coding model that assists people in writing, debugging, and reviewing code. Codex can work on a large, complicated project and even a speedy fix, and test its own code and find a bug before it creates a problem. Codex, according to OpenAI, provides more precise recommendations than previous ones, thus being a better ally to software teams.
The updates indicate the efforts of OpenAI to make ChatGPT a search alternative, as well as a practical tool for professional tasks.
Read Also : GSK Pledges $30 Billion US Investment to Boost Drug Research and Manufacturing
