On February 14, ByteDance officially rolled out the Doubao Large Model 2.0 series (Doubao-Seed-2.0), encompassing both the flagship Pro variant and the budget-friendly Lite version. The Doubao 2.0 Pro has attained a globally leading position in mathematics and reasoning skills, clinching gold medals in prestigious competitions such as the IMO and CMO mathematics contests, as well as the ICPC programming challenge. It has outperformed Gemini 3 Pro in the Putnam benchmark assessment. The model has also broadened its knowledge scope in niche areas, excelling in tests like SuperGPQA, with scientific domain knowledge scores on par with those of Gemini 3 Pro and GPT 5.2. Its multimodal comprehension abilities have undergone a comprehensive enhancement, achieving top-notch results in authoritative evaluations such as visual reasoning, spatial awareness, and long-context interpretation. Simultaneously, it has bolstered its support for time series and motion perception, facilitating real-time video stream analysis, environmental awareness, and proactive engagement, making it apt for life scenarios including fitness coaching, outfit recommendations, and caregiving companionship. Regarding Agent capabilities, Doubao 2.0 Pro has ascended to top-tier status in instruction adherence, tool utilization, and Search Agent evaluations, notably leading with a score of 54.2 in the HLE-Text test. Currently, Doubao 2.0 Pro is accessible on the Doubao App, computer client, and web version, where users can engage with it by opting for the "Expert" mode; Volcano Engine has also made API services available. In terms of pricing, Doubao 2.0 Pro bases its charges on input length, with inputs up to 32k incurring a cost of 3.2 yuan per million tokens and outputs costing 16 yuan per million tokens, presenting significant cost benefits over its rivals.