MiniMax-M3 debuts, eclipsing GPT-5.5 and Gemini 3.1 Pro on key benchmark performance for just 5-10% of the cost
Chinese AI startup MiniMax has launched its M3 large language model, which boasts advanced coding and multimodal capabilities at a significantly lower cost than leading proprietary models, starting at $20 per month. The model will be available under an open-source license, allowing for full customization and local deployment, potentially transforming enterprise AI infrastructure by enhancing data privacy and operational efficiency.
The release of MiniMax's M3 model, with its open weights and significantly lower cost compared to proprietary models, represents a major opportunity for enterprises to deploy high-performance AI locally while maintaining data privacy. The model's efficient MiniMax Sparse Attention mechanism reduces computational demands, offering a substantial cost advantage and enabling enterprises to customize the model deeply for specific applications. This makes M3 a highly attractive option for those looking to build autonomous software development or AI agent infrastructure without the constraints of closed-source systems.