OpenAI launches price war: GPT-5

OpenAI engages in price war: GPT-5.6 Luna plummets by 80%, and the price/performance ratio is positive DeepSeek V4 Pro The "price war" in the field of artificial intelligence has escalated again. On July 31, OpenAI officially announced the new pricing plan for the GPT-5.6 series, which significantly adjusted the prices of mid-range and lightweight models. The flagship model remained unchanged, while the lightweight model saw an unprecedented price reduction. According to the official plan, the Luna model for batch lightweight inference scenarios has become the core focus of this price reduction, and its input and output costs have been significantly reduced by 80% simultaneously. After the adjustment, the pricing per million input Tokens drops to US$0.2, while the pricing per million output Tokens is US$1.2. At the same time, the price of the Terra model of the same series, which focuses on balanced performance, has also been reduced by 20%, adjusted to an input of US$2 and an output of US$12 per million Tokens. However, the flagship version Sol, which focuses on complex code writing and multi-modal in-depth thinking, maintains its original high price, forming a clear product stratification strategy. Industry analysts pointed out that the unexpected sharp price reduction of OpenAI was largely due to the direct impact of the continued low-price offensive of domestic large models. Previously, domestic models represented by the DeepSeek V4 series achieved extremely low repeated input costs through innovative caching mechanisms. Conventional inference pricing during off-peak hours has remained at a low level for a long time, coupled with the affordable API services that many domestic manufacturers have continued to launch, which has continuously diverted overseas small and medium-sized developer customers. This has also forced OpenAI to adjust its pricing strategy for entry-level models. With the implementation of the new pricing plan, evaluations from third-party authoritative organizations have also been released. According to the evaluation chart released by Artificial Analysis, when completing the same standard tasks, the price-reduced GPT-5.6Luna has completely surpassed the DeepSeek V4Pro in terms of comprehensive cost performance. This change not only effectively makes up for the traditional shortcomings of overseas models in terms of cost performance, but also gives developers more competitive options in the selection of mainstream commercial models. via AI News (author: AI Base)