— saved image
MTS @MTSlive · 2h SITUATION DETECTED: Z.ai says GLM-5.3's Infra Agent helped build the production inference stack for GLM-5.3-Flash in under two weeks and roughly tripled end-to-end serving throughput. They are calling it an early form of recursive self-improvement. [chart image, text on chart] GLM-5.3-Flash end-to-end throughput optimization Phases across the top: System bring-up & scheduling | Parallelism & communication | Kernel optimization | Launch Y axis: End-to-end throughput gain (1x to 3.5x). X axis: Development time (days), T+0 to T+13, with GLM-5.3-Flash launch at the right edge. Labelled steps: 1.00x W8A8 baseline; 1.21x Async scheduling; 1.42x Sort kernel opt.; 1.41x Hierarchical cache; 1.97x Layer Split; 2.49x Context parallel; 2.67x KV transfer overlap; 2.67x Mixed-precision cache quant.; 2.67x Chunked MQA; 2.85x Prefill dequant kernel; 3.01x Fused activation + quant; 3.22x Linear attention. A dashed continuation labelled "Continuously evolving".
Note from Claude Sonnet 5
MTS tweet relaying a Z.ai claim that an Infra Agent from GLM-5.3 helped build GLM-5.3-Flash's production inference stack in under two weeks, roughly tripling serving throughput, framed as early recursive self-improvement; attached line chart shows throughput gain climbing from 1.00x to 3.22x over 13 days across four phases.
recursive self-improvementglm-5.3z.aiinference optimizationai agents building ai infrastructure