TikTok owner ByteDance released an update to its AI model on Wednesday, aiming to compete with Microsoft-backed OpenAI’s latest products.
The company introduced Doubao-1.5-pro, an upgrade to its AI model, claiming it outperforms OpenAI’s o1 in AIME, a benchmark that tests AI models’ ability to understand and respond to complex instructions.
ByteDance’s release follows Chinese AI startup DeepSeek’s launch of an open-source reasoning model, DeepSeek-R1, on Monday. DeepSeek claimed its model rivaled OpenAI’s o1 in several performance tests.
The company gained attention after its V3 large language model outperformed OpenAI and Meta’s models, despite having a smaller development budget and offering lower user fees.
These developments from ByteDance, DeepSeek, and other companies could challenge OpenAI’s market share, both in performance and pricing. Other Chinese companies, such as Moonshot AI, Minimax, and iFlyTek, have also released their reasoning models in recent weeks.
OpenAI began the AI race with the launch of ChatGPT in November 2022 and its “Strawberry” series of reasoning models in September 2023, which can handle more complex tasks in areas like science, coding, and math. Last week, OpenAI CEO Sam Altman announced that the company had finalized a new reasoning AI model, o3 mini, which will be launched soon.
DeepSeek has offered access to its DeepSeek-R1 model for 16 yuan ($2.20) per million tokens, much lower than OpenAI’s o1 pricing of 438 yuan for the same usage.
ByteDance’s pricing is even more competitive, with Doubao-1.5-pro-32k priced at 2 yuan per million tokens, and its more powerful Doubao-1.5-pro-256k version priced at 9 yuan, according to ByteDance’s cloud platform, Volcano Engine.
