OpenAI suddenly announced: GPT-5.6 has been significantly reduced in price, starting at 20% off

Zhidongxi Author | Jiang Yu Editor | Li Shuiqing Zhidongzhi reported on July 31 that today, OpenAI announced a reduction in the API price of the GPT-5.6 series model. Among them, the price of GPT-5.6 Luna has been reduced by 80%, the price of GPT-5.6 Terra has been reduced by 20%, and the GPT-5.6 Sol has a new Fast mode, which can achieve up to 2.5 times the inference speed while maintaining the same model capability. At the same time, OpenAI announced a price-performance distribution chart based on Artificial Analysis Intelligence Index v4.1. Judging from the figure, GPT-5.6 Luna is located in the upper left area of ​​the figure. While maintaining an intelligence index of over 50 points, the cost of a single task is approximately US$0.05 (approximately RMB 0.34 yuan). In comparison, Claude Opus 5, Gemini 3.6 Flash, GLM-5.2 Max and other models have almost similar intelligence index, but the cost of completing a single task is higher; DeepSeek V4 Pro is located further to the left, with a single task cost lower than most leading models, but the intelligence index is about 44 points. OpenAI hereby emphasizes that GPT-5.6 Luna can provide a better balance between cost and intelligence. ZXQKEEP10 ZXQ co-founder and CEO Sam Altman later issued a message saying that he hopes to "provide the best price/intelligence tradeoff at every model level." The day before, OpenAI had just disclosed that GPT-5.6 Sol had participated in optimizing its own inference system, reducing model service costs by 20%; optimizing the speculative decoding technology, and increasing Token generation efficiency by more than 15%; a day later, the benefits brought by this efficiency improvement were directly reflected in the product price. In order to more intuitively compare the market position after this price reduction,