ByteDance, the owner of the popular TikTok platform, has announced the launch of a new update to its artificial intelligence model, Doubao 1.5 Pro, which reportedly outperforms OpenAI’s o1 model in AI reasoning tests. This announcement comes as Chinese companies, such as DeepSeek, race to develop models capable of solving complex problems with AI reasoning capabilities, aiming to compete with global players such as OpenAI and others.
Read also: Alibaba cuts artificial intelligence price
Doubao 1.5 Pro: Enhancing logical reasoning
Representing a qualitative leap in ByteDance’s large language models, the new version Doubao 1.5 Pro comes as the company seeks to deliver artificial intelligence capable of understanding commands and answering multi-level questions with accuracy and flexibility. This update is a direct response to global market developments, especially following the announcement of OpenAI’s Strawberry series of models and the significant efforts made by companies in the United States and Europe to make models with AI reasoning capabilities and complex problem-solving available.
DeepSeek and open-source research
In parallel, DeepSeek announced the launch of its open-source DeepSeek-R1 model, which it claims matches or outperforms the performance of OpenAI’s o1 model in several standard benchmarks. This announcement has drawn the attention of industry experts globally, especially since DeepSeek previously introduced a model that achieved surprising performance despite lower funding than its competitors, pricing its services competitively compared to what global companies charge.
The open-source approach particularly attracts developers and researchers because it allows them to understand how the model works and contribute to its development, which could lead to faster and more collaborative technological breakthroughs.
Competitive pricing and challenging major corporations
ByteDance and DeepSeek are seeking to offer AI-powered services at fees much lower than those charged by global companies. For example, DeepSeek charges 16 yuan per million tokens, compared to 438 yuan for OpenAI’s o1 model. Meanwhile, ByteDance offers its Doubao 1.5 Pro 32K version at 2 yuan per million output tokens, and the higher Doubao 1.5 Pro 256K version at 9 yuan.
This pricing strategy puts pressure on international competitors and encourages Chinese companies and institutions to adopt local AI solutions. Given the size of the Chinese market and the enthusiasm of its institutions for AI process automation, this landscape could turn into a golden opportunity for a business model built on price and performance competitiveness.
Ongoing race in artificial intelligence
Other Chinese companies such as Moonshot AI, MiniMax, and iFlytek are joining the race to provide high-precision and fast logical reasoning models. This momentum reflects the high ambition characterizing China’s tech scene, where companies rush to pump massive investments into research centers and advanced infrastructure, driven by the desire to become a true competitor to Western tech companies.
It appears that reasoning capabilities in language models will be crucial for expanding into more complex practical applications, such as advanced cloud computing, digital education, intelligent robotics control, and other fields.
Implications for the global market
These developments could pose a genuine challenge to the dominance of OpenAI and other similar companies in the language model market, especially if Chinese companies continue to deliver high-quality models at low prices. This may push global companies to re-evaluate their pricing and investment policies, including moving toward opening parts of their systems or lowering the cost of access to AI services.
In the coming period, we are likely to see a series of alliances and acquisitions aimed at unifying efforts and narrowing research and development gaps. This competition may ultimately lead to a faster innovation environment and better services for users everywhere.
