Today, Li Liang, Vice President of Douyin, expressed his views on the rumors regarding ByteDance possibly initiating a price war for large models. He clearly stated on social media that this is not a price war, but rather a way to reduce the costs of using AI models through technological innovation.
It is reported that at the Volcano Engine Force Conference, ByteDance officially launched the Doubao visual understanding model, announcing its price at 0.003 yuan per thousand tokens input, which is 85% cheaper than the industry average.
In response to the rumors about ByteDance igniting a price war for large models, Li Liang stated that this is not a price war. He explained that the Doubao large model reduces costs through technological innovation, with significant optimizations made in algorithms, software engineering, and hardware solutions. The pricing of 0.3 cents per thousand tokens also allows for considerable gross margins. Furthermore, this is a straightforward transparent pricing model, not a "list price + discount" scheme. We hope to promote the widespread adoption and development of AI technology at a lower cost, as Tan Dai said: "A good model should be affordable for every enterprise."
It is worth mentioning that in light of the question regarding ByteDance's large models being cheaper than competitors, Tan Dai, President of Volcano Engine, also stated that the market needs sufficient competition, and that reducing costs is a result of technological optimization; only those who perform best can survive.
