List your company

Time-to-First-Token (TTFT) Latency on Controlled synthetic corpus Config A 1.0 (test)

57.9TTFT p50 LatencyAPC+Optimized
53.88880.969108.05135.131Jun 18, 2026

Evaluation results

MethodLinks
APC+OptimizedHardware=RTX 4060 Ti, Model=Qwen2.5-1.5B, Strategy=APC with optimized ordering (CacheWeaver)2026.0657.9161.368.118.9
APC+RetrievalHardware=RTX 4060 Ti, Model=Qwen2.5-1.5B, Strategy=APC with retrieval order2026.0672.5159.969.917.9
APC+SortedHardware=RTX 4060 Ti, Model=Qwen2.5-1.5B, Strategy=APC with lexicographic sorting2026.06108164.9101.725.3
No CacheHardware=RTX 4060 Ti, Model=Qwen2.5-1.5B, Strategy=Baseline2026.06158.2161.4158.30