MiniMax: MiniMax M3

MiniMax M3 from Minimax is designed to handle extensive text, image, and video inputs with a context length of 1048576 tokens. It supports tools and reasoning capabilities but does not offer structured output. The model excels in handling diverse input modalities and can reason effectively across multiple domains. This model is a solid choice for those needing robust processing power across text, images, and videos at a blended benchmark score of 78.0 from three independent evaluations. Its price point stands at $0.3 per million input tokens and $1.2 per million output tokens, making it competitive in terms of cost, especially for projects that heavily rely on diverse data types.