@XiaomiMiMo
π And the best part β we're open-sourcing it. 1,000+ tps on a 1T model wasn't a single breakthrough β it's deep model Γ system co-design between the MiMo and TileRT teams, all on general-purpose GPUs (no Cerebras-style wafer-scale, no Groq-style SRAM ASICs). On the model side: FP4 quantization (smaller footprint, less memory traffic) + DFlash, our block-masked parallel speculative decoding that accepts far more tokens per verification. On the system side, TileRT tailors its compiler & kernels to exactly these techniques. The result: a 1T model breaking 1,000 tps on a single, standard 8-GPU node. π€ Open weights (FP4 + DFlash checkpoint): https://t.co/jYQsgeruMg