Xiaomi: MiMo-V2-Flash (free)

MiMo-V2-Flash is an open-source foundation language model developed by Xiaomi, available for free on Multi AI. This advanced Mixture-of-Experts (MoE) model boasts 309 billion total parameters and 15 billion active parameters, utilizing a hybrid attention architecture for superior performance. It supports a hybrid-thinking toggle and an expansive 256K context window, making it highly effective for complex tasks. This model excels particularly in reasoning, coding, and agent scenarios. It has achieved the #1 ranking globally among open-source models on SWE-bench Verified and SWE-bench Multilingual, demonstrating performance on par with Claude Sonnet 4.5 while being significantly more cost-effective. For optimal and fastest performance when integrating with agentic tools like Claude Code or Roo Code, it is recommended to turn off reasoning mode. Key specifications include a 262K token context window and a 65K token maximum output. It offers streaming, functions, and long_context capabilities. Access this powerful AI model for free on Multi AI.