LMCache Storage ROI Calculator
RAM Space (Tier1) (GB):
Disk or Shared Storage Space (Tier2) (GB):
Model Type:
Big (Qwen3 coder 480B, Llama 70B)
Small (Qwen 30B, Llama 8B)
Sparse (Deepseek, GPT OSS)
GPU TCO Cost ($):
per hour:
or over 3 years:
TCO of 1 GB for tier 1 storage ($):
per hour:
or over 3 years:
TCO of 1 TB for tier 2 storage ($):
per hour:
or over 3 years:
(Cost per terabyte-hour. Your tier 2 volume is in GB). Source for default: S3 Express storage cost
Calculate
Break-Even Analysis
% of Reuses/Hours to Break Even:
Model Parameters
KV Cache Size:
Tokens/sec/GPU:
Prefill vs CacheBlend Speed Ratio:
Storage Calculations
Additional Storage Cost/Hour:
Additional Storage Cost/3y:
Processing Calculations
Number of Tokens cached:
GPU Hours to Prefill:
GPU Cost to Prefill:
GPU Cost for cache hit: