The Mach-Mind-4-Flash Technical Report introduces a 35B-parameter Mixture-of-Experts (MoE) agentic model with only 3B activated parameters. Achieved through post-training optimization without scaling pre-training compute, it performs on par with or surpasses 100B-parameter-class models.
The pipeline includes three key stages: a unified RL/OPD infrastructure delivering a 17% training speedup; Multi-Teacher On-Policy Distillation (MOPD) to fuse domain-specific experts; and Hybrid Median-length Policy Optimization (HMPO), which compresses reasoning chains by 19–46% with minimal accuracy loss.
The model leads or matches models with 10–30x its activated size on benchmarks like AIME'26 and IFBench while offering a fraction of the inference cost.