A：The non-IFRS financial measures discussed by management in this conference call are provided as supplemental information and are intended to aid in understanding, but are not a substitute for, the Company's financial results prepared in accordance with IFRS. The specific definitions of VIFR financial indicators and their correspondence to IFR financial performance were explained in detail at the meeting, and investors and analysts were reminded of the relevant risk factors.

A：The company's strategic focus in the next phase is to further consolidate the foundation, continue to build full-modal R & D capabilities, including language, video, voice, and music, and continuously improve products through technological innovation, such as an open platform for enterprises and developers, and Consumer-oriented intelligent assistants, etc. At the same time, the global layout will be strengthened, and the effect and performance of the M series model will be continuously tracked and improved to maintain its leading position in the global competition. In terms of language models, the company updated the M2, M2.1, M2 and M2.5 models in the fourth quarter of last year. Among them, M2 has been widely recognized by developers around the world for its outstanding performance in effect, price and speed balance, and has topped the global hot list. M2.1 focuses on improving the ability to deal with complex characters and instructions. M2 crane, as the underlying model, is committed to providing a more natural and personalized dialogue experience and has performed well in relevant tests. In addition, M2.5 economically achieves the cost feasibility of complex AD operations, with extremely fast reasoning, low operating costs, significant improvements in model capacity per generation, and rapid growth in model traffic.

A：In terms of multimodality, the company already has the ability to generate three modalities: video, voice and music. For example, the video model conch 2.3 has significantly improved the picture quality and stylization of human actions, while supporting a 50% reduction in batch creation costs. The speech model CPC2.6 optimizes the voice interaction experience, realizes the world's top ultra-low delay, and supports more than 40 languages. The music model has improved employment and can stably control various styles.

A：The company continues to integrate AI technology into its internal organization, covering more than 90% of its employees and multiple business scenarios. AI native products such as the M2 series model have entered the productization stage, and the minimax agent 2.0 has been launched, allowing users to create AI in professional fields locally. In addition, the company has also introduced Mac products to lower the threshold for users.

A：The company achieved $79 million in revenue for the full year of 2025, up 159 percent year-over-year, with AI native product revenue of $53 million, up 143 percent year-over-year, and open platform revenue of nearly $26 million, up 198 percent year-over-year. The international market accounts for more than 70% of the revenue, and the to B development platform products account for more than 50% of the revenue from the Chinese and foreign markets. Since the release of M2.5, it has gained significant attention in the international market, attracted many international customers to actively contact and cooperate, and has been deployed and used by global head cloud vendors and AI native platforms.

A：When we talk about platform companies in the AI era, its meaning is different from that of traffic entry platforms in the Internet era. Platform companies in the AI era will define and drive a new intelligent paradigm and enjoy paradigm dividends in products and businesses, which requires intelligent definition capabilities, technological innovation and product innovation capabilities, scalable infrastructure and efficient operation and maintenance capabilities.

A：We judge the following reasons: first, the AI market is not yet a stock game market, the annual increment is much greater than the stock, and innovative and unique companies can get opportunities. Second, in the next two to three years, we expect a significant increase in model development and infrastructure capabilities, and the possibility of opening up new application scenarios. In a high-quality market environment, we believe that we have the opportunity to gain advantages at the model level, product level and ecological level. Through long-term accumulation and rapid iteration, we have established high barriers at the model level. For example, we have launched multiple versions in the past 108 days and achieved rapid growth in the number of users. We have focused on the research and development of cross-modal models since the first day of our business, which gives us a unique advantage in the trend of multi-modal integration. In addition, minimax was the first company in China to develop both models and products, and this dual model-plus-product capability constitutes a strong barrier because model capability directly determines product capability, which is very unique and difficult to replicate.

A：At the ecological level, we have formed some small-scale ecology in open cloud and other fields by using the characteristics of our own model. Our model is widely used in open cloud platforms due to its high cost performance, lowering the threshold for developers and further lowering the threshold for users by integrating open cloud products (such as agent mass club).

A：R & D does face challenges, but it is necessary. From the first day of the company's start-up, we believe that the understanding of AGI should include dynamic input and output, so the different modalities in the organizational structure have the underlying ability to reuse. Under the organizational structure of AI native, although our full set of expenses is not high compared with other start-up companies, or even better than the investment of giants, there are already competitive models in each module. Our technical judgment and forward-looking verification in the past few years believe that this advantage will become more obvious in the future.

A：This is a very important issue. First of all, I explain what is the intelligence of L4 to L5 levels, as well as the possible future promotion direction and our position in it. We believe that the L3 level is the current normal autonomous driving technology, while the L4 and L5 levels are organizational intelligence. Taking our company as an example, an important mission is to develop more advanced models, which requires multi-person cooperation, involving comprehensive work such as algorithm innovation, experiments, training efficiency optimization, data processing and machine operation and maintenance. In such a development process, L4 level intelligence is reflected in the ability of a single researcher to complete innovative tasks, such as the design of an efficient solution based on a paper. The L5 level of intelligence means being able to effectively organize multiple people to work together. Regarding programming (coding), it is indeed part of the agent's capabilities, especially in productivity scenarios, such as the office field, where progress is fast and may be larger than the programming market.

A：What we see is a huge market. Although the programming field can help more people write code, there are not many people who really need to write code at work. Most of them are white-collar office workers. In big data analysis, financial report preparation, PPT production, etc., the number of professionals involved is far greater than the field of simple programming. We have made initial progress in the field of programming and IT, have unique advantages, and with the increase of resources, the model progress speed will accelerate and open a higher ceiling.

A：The first feature is the fast iteration speed. The M2 series model has achieved significant improvement in only 108 days, which proves our R & D capability and model carrying capacity. The second feature is the ability to create unique models, such as M2 Conch II and CB II, to differentiate and open up market space by defining unique advantages.

A：We are striving to become a platform company in the AI era. The core driving force is the continuous improvement of intelligent density and large token throughput. in terms of strategic positioning, we insist on all-mode capability enhancement and unique value creation, and focus resources on model research and development and product innovation that can generate unique value, such as agent, transparent beverage products and conch video. At the same time, we insisted on multi-modal accumulation from the beginning and kept ahead in research and development efficiency, making full use of agile organizational forms and experience reuse to maximize the speed of model progress and rapid growth of commercialization revenue. We believe that in the global scope, there will only be a few AI platform companies that can lead the industry, and we have certain advantages and uniqueness, and we are expected to become one of them. For the current growth rate, although there are expectations, it is still amazing. It may be related to market demand, technological upgrading and other factors. This trend tends to be a long-term sustainable process rather than a short-term one-off dividend.

A：This internship program, which covers 90% of the employees, allows us to plan the company's development in a forward-looking time, and through the extensive use of internal AI technology, significantly improves organizational efficiency. We are working hard to build the company into a AI native organization. Through AI empowerment, we have accelerated the entire closed loop of model iteration, product innovation and service users, enabling us to define and achieve model goals faster and allow AI to maximize their value within the company.

A：The value of the company's internal AI model will become more significant as the iteration speed, revenue growth, service user capacity and throughput capacity continue to improve. When these models are running within the company, we can clearly observe that even the best models at present still need to be improved in many aspects, and these problems that have not been completely solved have the highest economic value and use value. They will directly guide the research and development direction of our next generation models and ADC, help us define our goals more quickly and clearly, and form a positive flywheel, Become the company's continuous core competitiveness.