OpenAI officially announced that the default model of the ChatGPT chatbot has been upgraded to GPT-5.5Instant. This upgrade focuses on improving the accuracy and conciseness of the model, especially in interactions with users, by reducing unnecessary elements in responses, such as emojis. The new model aims to provide a more efficient communication experience.
GPT-5.5Instant has optimized its response content, aiming to further reduce verbose expressions while maintaining practicality. Users will experience more intuitive and clear responses, which is particularly important for daily communication. OpenAI stated that this upgrade not only enhances the performance of the chatbot but also reduces the "hallucination" phenomenon when providing information, ensuring users receive more reliable information.
According to OpenAI's internal evaluation, the "hallucination" statements of GPT-5.5Instant have decreased by 52.5%. In tests where users marked conversations with factual errors, inaccurate statements decreased by 37.3%. This means that in high-risk fields such as healthcare, law, and finance, GPT-5.5Instant provides a significant improvement in accuracy. OpenAI noted that the Instant series has achieved comprehensive progress in factual accuracy, with the greatest benefits in areas that require the highest level of accuracy.
This update is now available to all users, while paying users can continue using the previous GPT-5.3Instant version for a three-month transition period. This move aims to ensure users have enough time to adapt to the new version and smoothly transition to a more advanced model.
Key points:
🌟 GPT-5.5Instant is OpenAI's latest released chatbot model, emphasizing conciseness and accuracy.
📉 Hallucination statements reduced by 52.5%, improving information accuracy in high-risk fields.
🗓️ All users have started the update, and paying users can continue using the old version for three months.
