MiniMax Intelligence Launches Upgraded SuperMate On-Device Intelligent Cockpit Solution

Introduction: On-Device Large Models Enter a New Era for Automotive Cockpits

On April 24, the 19th Beijing International Automotive Exhibition (Auto China 2025) opened its doors. At this globally watched automotive industry event, AI large model company MiniMax Intelligence officially released the upgraded version of its SuperMate on-device intelligent cockpit solution, drawing widespread industry attention. The launch not only demonstrates the deep application potential of on-device large model technology in automotive scenarios but also signals that the integration of AI and the automotive industry is advancing from proof-of-concept to a new phase of scaled mass-production delivery.

Notably, SuperMate has already been deployed in mass-production vehicles from automakers including Geely and Changan Mazda, placing the company in a leading position in the race to bring on-device large models into vehicles.

Core: SuperMate Upgrade Redefines Cockpit Experience with On-Device AI

The newly upgraded SuperMate released by MiniMax Intelligence represents the latest achievement of its sustained efforts in the on-device intelligent cockpit domain. Unlike traditional intelligent cockpit solutions that rely on cloud computing power, SuperMate's core advantage lies in deploying large model capabilities directly on in-vehicle chips, delivering a "localized, low-latency, high-privacy" intelligent interaction experience.

The term "on-device" means the large model performs inference directly on the terminal device (in-vehicle chip) without relying on cloud servers. This technical approach delivers three significant advantages:

First, dramatically improved response speed. Since model inference is completed locally, users' voice commands and interaction requests can be processed at the millisecond level, avoiding the latency issues associated with cloud communication. In driving scenarios, this instant response capability is particularly critical, directly impacting user experience and driving safety.

Second, more reliable privacy protection. On-device deployment means users' voice data, behavioral habits, and other sensitive information do not need to be uploaded to the cloud, fundamentally reducing the risk of data breaches. Against the backdrop of increasingly stringent global data privacy regulations, this feature carries significant compliance value.

Third, significantly enhanced offline availability. Even when vehicles travel through areas with poor signal coverage — such as tunnels, mountainous regions, or underground parking garages — SuperMate can maintain full intelligent interaction capabilities without "downgrading" to a traditional infotainment system due to network interruptions.

MiniMax Intelligence is known to have deep technical expertise in efficient inference and model compression for on-device large models. Its independently developed MiniCPM series of small-parameter models maintains powerful language understanding and generation capabilities while dramatically reducing computational resource requirements, making it possible to run large models smoothly on automotive-grade chips. The latest SuperMate upgrade delivers notable improvements in model capability, interaction fluidity, and scenario coverage.

Analysis: Mass Production Becomes the Key Watershed as MiniMax Intelligence Seizes First-Mover Advantage

In the current intelligent cockpit landscape, "bringing large models into vehicles" has become an industry consensus, yet companies that have truly achieved mass-production delivery remain few. MiniMax Intelligence's SuperMate solution has been deployed in multiple mass-production models from well-known automakers including Geely and Changan Mazda — a milestone achievement that serves as an industry benchmark.

From an industry structure perspective, intelligent cockpits are undergoing a profound transformation from "feature stacking" to "intelligence emergence." Over the past few years, automakers have continuously added large screens, voice assistants, gesture recognition, and other features to cockpits, but user experiences have often been fragmented. The introduction of large models promises to fundamentally change this situation — through powerful natural language understanding and multi-turn dialogue capabilities, various cockpit functions can be integrated into a unified intelligent interaction portal, allowing users to communicate with vehicles in more natural ways.

However, the challenges of bringing large models into vehicles cannot be overlooked. The computational constraints of automotive-grade chips, the high real-time performance requirements of cockpit scenarios, the power consumption and thermal management limitations of in-vehicle deployment, and automakers' strict demands for supply chain stability and cost control all set an extremely high bar for technical solutions. MiniMax Intelligence's ability to achieve mass-production delivery across multiple mainstream automakers demonstrates that it has developed a relatively mature technology system and delivery capability in model efficiency optimization, automotive-grade adaptation, and engineering deployment.

Looking at the choice of partners, Geely and Changan Mazda represent typical examples of Chinese independent brands and joint-venture brands respectively. SuperMate's ability to gain recognition from both types of automakers reflects the solution's comprehensive strength in technical versatility and commercial maturity. This also lays a solid foundation for MiniMax Intelligence to expand partnerships with more automakers going forward.

Furthermore, from a supply chain perspective, the trend of on-device large models entering vehicles will also have far-reaching implications for upstream chip manufacturers, middleware suppliers, and downstream user service ecosystems. As more vehicle models are equipped with on-device AI capabilities, the application ecosystem surrounding intelligent cockpits is expected to accelerate its growth, forming a complete value chain of "chips — models — applications — services."

Outlook: On-Device AI Set to Become Standard in Intelligent Vehicles

Looking ahead, the application prospects for on-device large models in the automotive sector are vast. As automotive-grade chip computing power continues to improve and model compression and quantization technologies achieve ongoing breakthroughs, the capability boundaries of on-device AI will continue to expand — progressing from current voice interaction and information queries to deeper intelligent scenarios including multimodal perception, proactive driving assistance recommendations, and personalized service suggestions.

For MiniMax Intelligence, the release of the upgraded SuperMate is merely a new starting point. Building on existing mass-production partnerships, how to further expand its automaker collaboration portfolio, deepen scenario coverage capabilities, and build an open cockpit AI ecosystem will be key challenges for its future development.

From a broader perspective, 2025 is becoming the "acceleration year" for deep integration between AI large models and the automotive industry. At this year's Beijing Auto Show, multiple companies showcased their respective AI cockpit solutions, and competition is intensifying. What is certain, however, is that on-device intelligence will gradually evolve from a "differentiating selling point" to a "foundational standard" for intelligent vehicles, and companies that are first to achieve scaled mass-production delivery will occupy the most advantageous positions in this transformation.

With SuperMate, MiniMax Intelligence has proven one thing: bringing on-device large models into vehicles is no longer a laboratory technology demonstration — it is a reality that is driving into households everywhere.