Xiaomi MiMo-V2.6 Series Open-Source Multimodal Reasoning Models Now Available

Key Info

Xiaomi MiMo-V2.6 series open-source multimodal reasoning models are now available on the B.AI platform, featuring a 1M token context window and full text/image/video/audio understanding.

Highlights

• MiMo-V2.6-Pro: flagship sparse MoE with 1.02T total / 42B active parameters, for repository-level software engineering, long-horizon agents, and complex multimodal reasoning.
• MiMo-V2.6-Flash: efficient 309B total / 15B active parameters, for high-throughput office automation and large-scale agent calls.
• Both models are accessible via API and Web Chat on the platform; official documentation links provided.