Twitxy Saved

Tensormesh

Verified account This account is verified. Learn more

@tensormesh

AI inference optimization company that never charges you twice for cached tokens, making AI applications faster and dramatically cheaper to run anywhere.
tensormesh.ai Joined October 2025
24 Following 375 Followers 183 Posts

Latest posts 5

Tensormesh Verified account @tensormesh
Running LMCache across a fleet? Each KV cache server only knows about itself, so the same prompt gets prefilled again on every node.

The LMCache Coordinator adds a control plane on top: ๐จ๐ง๐ž ๐Š๐• ๐œ๐š๐œ๐ก๐ž ๐Ÿ๐จ๐ซ ๐ญ๐ก๐ž ๐ฐ๐ก๐จ๐ฅ๐ž ๐Ÿ๐ฅ๐ž๐ž๐ญ.

Itโ€™s the cross-node layer in Tensormesh Platform, our self-hosted KV caching for vLLM on Kubernetes ๐Ÿ‘‡
na2.hubs.ly/H08-ygm0
Tensormesh Verified account @tensormesh
Prefix caching stops at the first change in your prompt. RAG and agents reorder the same docs on every call.

๐‚๐š๐œ๐ก๐ž๐๐ฅ๐ž๐ง๐ reuses KV cache anywhere in the prompt:
- 2.6x more traffic, same GPUs
- Up to 2.9x faster TTFT
- ~$2.7M/yr saved on 256 GPUs
na2.hubs.ly/H08VHzK0 @CloudDude_
tensormesh.ai CacheBlend: 2.6x More RAG Traffic, Same GPUs | Tensormesh Non-prefix KV cache reuse: CacheBlend carries 2.6x more RAG traffic on the same GPUs. See the benchmark and estimate your savings.
Tensormesh Verified account @tensormesh
Today at #KubeCon โฐ 1:30 PM ยท @SUSE booth #107

Sandro Mazziotta on sovereign AI inference at a predictable cost. ๐Ÿ’๐Ÿ“ ๐ฆ๐ข๐ง๐ฎ๐ญ๐ž๐ฌ, ๐›๐ซ๐ข๐ง๐  ๐ฒ๐จ๐ฎ๐ซ ๐ช๐ฎ๐ž๐ฌ๐ญ๐ข๐จ๐ง๐ฌ.

#SUSE #SUSEAIFactorywithNVIDIA