Xiaomi's First Fully Multimodal Large Model Debuts: The Final Piece in the "People-Vehicle-Home" Multi-Scenario Strategy?

Since Xiaomi released its self-developed AI large model Xiaomi MiMo-V2-Flash at the 2025 December All-Ecosystem Partner Conference for Mobility, Vehicles, and Home, Xiaomi has accelerated again.

On March 19, Xiaomi announced its first all-modal foundation model, Xiaomi MiMo-V2-Omni.

MiMo-V2-Omni is designed as an “executor” with cross-modal perception and GUI (Graphical User Interface) operation capabilities, seamlessly integrating with various agent frameworks.

Previously, the model was blind-tested under the code name “Healer Alpha” on the OpenRouter platform and demonstrated performance that matches or even surpasses some leading closed-source models in various benchmarks.

Regarding the model’s efficient “speed of innovation,” Lei Jun stated: “We are relatively low-key in the AI field, and our actual progress may be much faster than what everyone sees. In AI, our R&D and capital investment this year will exceed 16 billion yuan. I believe that as long as we continue to invest steadily, Xiaomi will deliver a brilliant result in the AI era.”

Furong Luo, the core person responsible for this model, also openly said on overseas social platforms: “Before tomorrow, anyone in the MiMo team who has conducted fewer than 100 dialogue tests can leave immediately. This move has been effective. Once the team’s imagination is ignited by the capabilities of the intelligent system, that imagination directly translates into R&D speed.”

Currently, Xiaomi offers API pricing at $0.4 per million tokens for input and $2 per million tokens for output (supporting 256K context).

Xiaomi’s ambitions clearly go beyond just selling APIs to developers.

The model has already partnered with Kingsoft Office (WPS) to explore scenarios involving text generation and structured data processing.

From a strategic perspective, the ultimate goal of MiMo-V2-Omni points toward Xiaomi’s “full ecosystem of mobility, vehicles, and home.”

In future visions for MiMo-V2-Omni, Xiaomi also states it will “continue to promote long-term intelligent agent planning, real-time streaming perception, multi-agent collaboration, and deeper integration with the physical world.”

If this model can be deeply integrated as the underlying “brain” into Xiaomi’s HyperOS, creating an AI foundation capable of cross-device deep understanding of voice commands, autonomous app invocation, and even controlling Xiaomi vehicle interfaces, it would significantly enhance Xiaomi hardware’s premium value and user retention.

Despite the compelling technological demonstrations and ecosystem visions, Xiaomi currently faces severe challenges in resource allocation and cost control.

At present, Xiaomi is operating under high pressure with multiple fronts:

On one hand, its cash cow smartphone business is facing headwinds from soaring upstream storage chip prices, squeezing overall hardware gross margins; on the other hand, its automotive business is in a critical phase of capacity ramp-up and nationwide sales network expansion, requiring continuous investment.

Moreover, compared to pure internet giants with substantial profit margins and large cloud computing bases, Xiaomi’s funding in the AI arms race is less advantageous.

From a strategic vision, MiMo-V2-Omni is undoubtedly the most critical piece for Xiaomi to complete its “full ecosystem of mobility, vehicles, and home” intelligent loop.

In the context of rising memory prices, balancing multi-line investments in smartphones, automobiles, and large models tests Xiaomi’s management wisdom.

Risk Warning and Disclaimer

The market carries risks; investments should be cautious. This article does not constitute personal investment advice and does not consider individual users’ specific investment goals, financial situations, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular circumstances. Invest accordingly at your own risk.