MiniMax发布新模型M2.5，1美元/小时

Core Viewpoint - The domestic AI industry is experiencing a surge in model releases, with MiniMax announcing its new text model, MiniMax M2.5, which emphasizes cost-effective programming capabilities [2] Group 1: Model Features and Economic Impact - MiniMax M2.5 can output 100 tokens per second for a cost of $1 per hour, and at 50 tokens per second, the cost drops to $0.3 per hour, indicating a significant reduction in operational costs for AI agents [2] - The company believes that when performance and cost are no longer constraints, the economic model for scaling agent deployment will fundamentally change [2] Group 2: Internal Implementation and Performance - MiniMax has fully adopted an "AI-assisted" mode internally, with 30% of overall tasks being autonomously completed by M2.5 across various functions including R&D, product, sales, HR, and finance [2] - The model has shown exceptional performance in programming tasks, generating 80% of new submitted code, highlighting its effectiveness in real business scenarios [2]