Recently, ByteDance’s large model pricing strategy has attracted industry attention and has been interpreted as potentially triggering a price war. In response to this, Douyin Vice President Li Liang responded on social media. He denied that this was a price war and explained in detail ByteDance’s strategy to reduce the cost of using large models.
Today, Douyin Vice President Li Liang expressed his views on rumors that ByteDance may start a price war for large models. He made it clear on social media that this is not a price war, but a reduction in the cost of using AI models through technological innovation.
It is understood that at the Volcano Engine Force conference, ByteDance officially released the Beanbao visual understanding model, and also announced that its price is 0.003 yuan per thousand tokens, which is 85% cheaper than the industry average price.
In response to rumors that ByteDance is launching a price war for large models, Li Liang, vice president of Douyin, wrote in response: This is not a price war. He said that Doubao Big Model reduces costs through technological innovation and has made many optimizations in algorithms, software engineering and hardware solutions. The pricing of 3 cents/thousand tokens also has considerable gross profit. And this is a one-step transparent price, not a "regular price + discount" method. We hope to use lower costs to promote the inclusive development and application development of AI technology. As Tan Dai said: "A good model is one that every enterprise can afford to use."
It is worth mentioning that due to the problem that large byte-series models are cheaper than their peers, Volcano Engine President Tan Dai also said that the market needs full competition, and cost reduction is the result of technological optimization. Only by doing the best can we survive.
ByteDance’s move aims to reduce the cost of using AI models through technological innovation and promote the inclusive development of AI technology, rather than pure price competition. Its transparent pricing strategy and emphasis on technology optimization reflect its long-term development strategy and understanding of market competition. The future direction of the large model market still requires continued attention.