# Long-term memory for AI

The authors release the public package galahad-kv, which writes the KV state a model computes for each 16,000-token block to encrypted local NVMe and later restores any block byte-exact with zero recompute; on a 50-million-token real corpus served through vLLM on one NVIDIA H100, all 100 probed blocks were restored without recompute for both Gemma 4 12B and 31B, restores were 2.8–4.3× faster and used 8.8–12.3× less GPU energy than recompute, VRAM stayed flat, and planted-fact recall was 82/100 and 98/100 with zero fabrication.

The authors release the public package galahad-kv, which writes the KV state a model computes for each 16,000-token block to encrypted local NVMe and later restores any block byte-exact with zero recompute; on a 50-million-token real corpus served through vLLM on one NVIDIA H100, all 100 probed blocks were restored without recompute for both Gemma 4 12B and 31B, restores were 2.8–4.3× faster and used 8.8–12.3× less GPU energy than recompute, VRAM stayed flat, and planted-fact recall was 82/100 and 98/100 with zero fabrication.