FULL STORY

Redis Author Optimizes DeepSeek V4

Redis author antirez shared progress on the DwarfStar engine, significantly boosting DeepSeek V4 inference speeds to 170 t/s via quantization and routing.

2026-08-17 ~ 2026-08-17 · 2 episodes · 5 posts

Episode 1 · Redis Author Optimizes DeepSeek V4 to 45 t/s (2026-08-17, 3 posts)

Redis author antirez demonstrated the DwarfStar inference engine, optimizing DeepSeek V4 PRO to 45 t/s via Q2 quantization and dynamic expert routing. The system adaptively migrates experts between VRAM and RAM based on usage.

Episode 2 · antirez Tunes DwarfStar to 170 t/s Generation on Station (2026-08-17, 2 posts)

Redis creator antirez reported that after roughly 48 hours of optimization, the DwarfStar inference engine on Station achieved 170 t/s generation (without DFlash) and 22k tokens/s prefill using MXFP4 and Flash models.