MiniMax M3.1 Flash Released with Open Model Weights
MiniMax has officially released its M3.1 Flash model, alongside the open-source model weights. This multimodal release marks a notable update from the previous M3 version. The availability of open model weights allows developers and researchers in the local AI community to run, customize, and inspect the model locally. This fosters greater accessibility and innovation within the open-weight AI ecosystem. Compared to its predecessor M3, MiniMax M3.1 Flash delivers significant improvements across multiple benchmarks, particularly excelling in complex agentic tasks, coding, and cybersecurity.
## BACKGROUND
MiniMax is an artificial intelligence company known for developing advanced multimodal and language models. Open model weights refer to making the internal parameters of an AI model public, allowing users to host and fine-tune the technology independently rather than relying solely on cloud-based APIs.