BAAI says PhysisServe, evaluated across five ACWM and WAM backbones with full weights and no pruning, compression, or quantization, reaches 4.1 ms single-stream output latency for the 33B MiniMax H3 with one GPU per stage.
The post says this equals about 245 chunks per second, or roughly 1,000 FPS as an ACWM and up to 3,900 Hz as a WAM. The evaluated backbones are Wan2.1 1.3B, Wan2.2 TI2V 5B, Wan2.1 14B, LTX-2.5 22B, and MiniMax H3 33B.
Part of a thread · earlier post
Benchmark Results: We evaluated PhysisServe across 5 widely used ACWM and WAM backbones—with full weights and zero pruning, compression, or quantization:
Evaluated Backbones: Wan2.1 1.3B, Wan2.2 TI2V 5B, Wan2.1 14B, LTX-2.5 22B, and MiniMax H3 33B.
Ultra-Low Latency: With one GPU per stage, the33B MiniMax H3achieves a single-stream output latency of just4.1 ms(~245 chunks/s).
Real-Time Throughput: Translates to~1,000 FPSas an ACWM and up to3,900 Hzas a WAM!
Source: BAAI · x.comPublished · added here
