Compatibility matrix
Tensormesh is released as a matched set. Deploy a single row’s versions together — the cache engine, operator, and Helm chart are supported as a unit.
Current as of 2026-10-19.
What a release ships
Every artifact that makes up this release, and where to pull each one. Install them as a set — a Helm chart from one release with an operator image from another is not a supported combination.
Manifest for release bundle
lm-v0.5.6_op-v1.0.7_ch-1.0.1.
Every reference above except the cache engine is served through
artifacts.tensormesh.ai and needs your API key. The cache engine image is
public and is pulled straight from Docker Hub, so an air-gapped install must
mirror that one separately.Model Support
In Tensormesh Platform we distinguish what is verified and supported. A verified model or feature means it is part and tested as part of Tensormesh CI. It is ready for production. A supported model or feature requires testing prior to go on production as it is supported on LMCache opensource version but it has not been verified in our CI. In other word, we know/assume it works but it is not part of our CI process.Supported models
CPU/Storage offloading
CacheBlend
Tensormesh Platform Verified
CPU offloading performance
CPU offloading performance speed up verified in the current release.Cache features & External storage
Cache and storage capabilities available on the current release.On the roadmap: GPUDirect Storage (GDS), Peer-to-peer KV transfer.
CacheBlend
Non-prefix KV cache reuse — cached chunks are reused at shifted positions (re-RoPE + partial recompute), beyond what prefix caching can do.Serving engines
Serving engines supported in front of the Tensormesh cache.On the roadmap: TensorRT-LLM, vLLM via stack-operator.