Non-prefix KV cache reuse for RAG & agents: CacheBlend-style selective recompute, provenance-keyed cache, reuse gate, and an honest benchmark (MLX + PyTorch/CUDA)
benchmark pytorch mlx rag kv-cache apple-silicon llm-serving vllm llm-inference qwen llm-agents prefix-caching lmcache cacheblend
-
Updated
Oct 24, 2026 - Python