OpenAI releases GPT-5.1: Intelligence and Emotional Quotient Improved
On November 12 local time, OpenAI officially announced the launch of its new flagship model series, GPT-5.1. The official stated that this upgrade aims to "make ChatGPT smarter and the conversation experience more interesting." This release came just a few months after the previous generation, GPT-5. Previously, GPT-5 had sparked user controversy due to its lack of "human touch" and being "rigid and lacking creativity." In contrast, GPT-5.1 has achieved dual upgrades in intelligence level and communication style.
According to the official announcement, the GPT-5.1 series includes two core models: GPT-5.1 Instant and GPT-5.1 Thinking. It also introduced the GPT-5.1 Auto intelligent feature, which will be gradually rolled out to paying users starting today. Free users will follow after the coverage of paying users is completed. API developers can access GPT-5.1 later this week. To avoid user adaptation issues, OpenAI also provided a "regret remedy," allowing paying users to switch back to the previous version of GPT-5 within three months through the settings, ensuring a smooth transition.
As the most commonly used default model for ChatGPT, GPT-5.1 Instant adds an "adaptive reasoning" mechanism, becoming the first lightweight model to introduce a chain of thought (CoT). It can autonomously judge the complexity of questions, maintaining ultra-fast response for simple queries and automatically engaging in deep thinking for complex tasks. At the same time, its tone is more friendly and talkative, with occasional playful traits making conversations more interesting. The accuracy of instruction following has also been significantly improved, achieving "fast and accurate, warm and practical."OpenAI releases GPT-5.1: Intelligence and Emotional Quotient Improved
Original: www.toutiao.com/article/1848636771521732/
Statement: This article represents the views of the author.