Corbenic AI launches Galahad beta to reuse GPU cache on SSD
Belgian startup Corbenic AI launched its Galahad beta software on October 1, 2026, to allow servers to reuse encrypted key-value caches from local NVMe drives.
Galahad saves model cache
The new tool stores a model's key-value attention state on encrypted local storage. This allows the system to skip rereading text after GPU memory is cleared or following a restart.
Speedup comes from less reading
Founder Sietse Schelpe told TechRadar Pro that no special hardware is required for this process. The company credits most of the reported 14x speedup to Blaise, a retrieval component rather than SSD loading speeds.
Existing tools persist cache
Other solutions like LMCache and Nvidia Dynamo already persist KV cache to storage systems. Galahad distinguishes itself by offering exact, fail-closed reuse within a single small library for vLLM, SGLang, and Llama.cpp.
Reported by one outlet
Only one outlet has published this. Nothing here has been checked against a second report, so read it as that outlet's account and follow the link below for the original.
Reported by
1 independent outlet. Headline as published. Links open the original report.