MiniMax unveils M3 AI model for complex coding tasks

Chinese AI startup MiniMax has launched its M3 flagship model, designed for coding agents and automated workflows. The Shanghai-based company announced on Monday that the model's architecture reduces computational requirements to one-twentieth of previous levels, slashing inference costs and boosting response speeds. M3 can process up to 1 million tokens at once, five times more than its predecessor M2.7, enabling it to handle long, complex programming projects. In a benchmark test, M3 optimized software running on Nvidia's Hopper chips. According to MiniMax's WeChat post, the model outperformed OpenAI's GPT-5.5 and Google's Gemini 3.1 Pro on the SWE-Bench Pro coding benchmark. The company did not disclose the model's size or training infrastructure.