MiniMax first disclosed its performance! It will compete with Zhipu for the title of "first stock" in large models

Following Zhipu, MiniMax (Xiyu Technology) disclosed its post-hearing information set for the first time on December 21.

Founded in early 2022, MiniMax is a general artificial intelligence technology company that insists on self-research across text, video, and voice modalities. It is expected to set a record as the AI company with the shortest time from establishment to IPO.

Its team is also highly youthful, with founder Yan Junjie born in 1989, previously serving as Vice President of SenseTime, Deputy Dean of the Research Institute, and CTO of the Smart City Business Group, leading the development of deep learning toolchains and general intelligent technology systems. As of the end of September 2025, the company has 385 employees, with an average age of 29, of which nearly 74% are R&D personnel, and the average age of the board members is 32.

As two major "unicorns" in the large model field, it remains uncertain who will cross the finish line first between Zhipu and MiniMax to become the "first stock." However, some executives from large model companies believe that while Zhipu and MiniMax are both leading enterprises and are advancing towards listing, there are differences in core dimensions such as technical routes and business models, each with its competitive track. After their listings, they will fill the gap in the capital market for large model targets and provide valuation references for the industry in the secondary market.

Business Perspective Advocating "Model as Product"

The post-hearing information set shows that MiniMax's revenue for the first three quarters of 2023, 2024, and 2025 was $3.46 million, $30.523 million, and $53.437 million, respectively, with a year-on-year growth rate of 782.2% for the full year of 2024 compared to 2023, and a year-on-year growth rate of 174.7% for the first nine months of 2025; overseas revenue accounted for approximately 19.2%, 69.8%, and 73.1%, respectively; adjusted net losses were $89.074 million, $244 million, and $186 million.

In terms of commercialization path, MiniMax adopts a relatively rare approach among startups—"model as product," advancing both the model and its applications simultaneously. Yan Junjie has stated: "In the era of large models, the real product is actually the model itself; traditional products are more like a channel. If the model is not smart enough, it doesn't matter how well the product is made."

Therefore, MiniMax is more like an "AI product factory." Currently, MiniMax has launched a series of AI-native applications for the global market, including Hai Luo AI, Xing Ye, Talkie, and an open platform for enterprises and developers

Conch AI Generation

Among them, the video production product Conch AI is MiniMax's most "outstanding" product, with the "Animal Diving" video generated by this application going viral on overseas social platforms, with views exceeding 300 million.

Specifically, in terms of business, MiniMax's business model is primarily based on global paid subscriptions and model invocation revenue, divided into To C and To B businesses. The B side does not involve privatization deployment business, pursuing recurring revenue, which is currently growing rapidly, and overseas revenue has far exceeded domestic revenue, accounting for over 70% of total revenue; the C side is also selectively engaged, not participating in dialogue product competition, focusing on content tools, content platforms, agents, and other directions.

On the B side, MiniMax entered the voice and music generation business relatively early, embedding voice capabilities into audiobooks, AI toys, digital humans, humanoid robots, etc., with clients including Xiaomi, Yuewen, Gaotu, Fuzozo, and Zhiyuan Robotics. On the C side, the AI multimodal interaction platform Talkie and the video generation product Conch AI have brought stable cash flow to MiniMax.

From a business perspective, Conch AI, Talkie, and Starry Sky are important supports for MiniMax's revenue, contributing revenues of $17.464 million and $18.75 million respectively in the first three quarters of 2025, accounting for 32.6% and 35.1% of total revenue; the average spending per paying customer is $56 and $5 respectively.

In addition, revenue from "open platforms and other AI-based enterprise services" has seen significant growth, contributing $15.417 million in the first three quarters of 2025, a year-on-year increase of 160%, accounting for 28.9% of total revenue; the average spending per paying customer is $6,167.

In terms of gross margin, MiniMax quickly turned positive from -24.7% in 2023, reaching 12.2% in 2024, and further improving to 23.3% in the first nine months of 2025. The gross margin for the C side is 4.7%, with the overall gross margin of the company close to 50% excluding Starry Sky.

Technological Advancement in Three Major Modalities

Compared to domestic and foreign large model companies, the technical route followed by MiniMax is somewhat "non-mainstream"—the industry generally believes that one should first focus on achieving excellence in a single modality such as text before advancing to integration, thereby achieving AGI; MiniMax believes that achieving AGI (Artificial General Intelligence) is a systematic project, akin to building rockets and chips, with each link closely interconnected Therefore, MiniMax did not adopt a model of deepening a single modality before integration, but instead bet on the self-research of text, vision, and audio modalities from day one. For a startup, synchronously advancing multiple modalities also means high investment and high risk, which can be somewhat aggressive.

In this regard, individuals familiar with MiniMax explained that 80% of the resources across modalities can be shared, and it is precisely because the three major modalities were advanced simultaneously from day one that the current push for modality integration is "much easier" than for other companies.

At the model level, MiniMax is currently one of the few companies globally that has entered the first tier in all three major modalities. Its text model MiniMax M2 ranks in the top five globally on the authoritative evaluation list Artificial Analysis, and is the first in open source; the video model Hailuo 02 series has set a record for the cost-effectiveness of global video models, ranking in the top three on multiple lists; the speech model Speech02 series has surpassed OpenAI on international authoritative evaluation lists, reaching the top position globally.

MiniMax does not build its own computing power but rents cloud computing power for training and inference. Therefore, the company includes cloud computing service costs related to inference in its sales costs. Including platform commission fees, employee costs, etc., MiniMax's sales costs for the first three quarters of 2023, 2024, and 2025 were $4.314 million, $26.785 million, and $40.961 million, respectively.

Cloud computing service fees related to training are included in R&D investment. After accounting for employee costs, stock compensation, and other expenses, MiniMax's R&D investments during the period were $70.002 million, $189 million, and $180 million, respectively.

From its establishment until September 2025, MiniMax has spent a total of $500 million (approximately 3.5 billion RMB). In comparison, OpenAI's cumulative expenditure ranges from $40 billion to $55 billion, meaning MiniMax has spent less than 1% of that amount.

Since its establishment, MiniMax has raised over $1.5 billion in financing, with strategic investors including miHoYo, Alibaba, Tencent, Xiaohongshu, Xiaomi, Kingsoft, PCG, and Charoen Pokphand Group; financial investors include Hillhouse Capital, IDG, Sequoia, Matrix Partners, Ming Shi, and Yunqi Capital.

As of the end of September 2025, MiniMax's cash balance was $1.046 billion. Based on the current cash burn rate forecast, even without IPO fundraising, the company's cash is sufficient to support its operations for over 37 months Shanghai Securities Journal

Risk Warning and Disclaimer

The market carries risks, and investments should be made with caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this is at one's own risk