Notes from the team.
Announcements, product releases, and research notes from the RunInfra team.
With software alone, one B200 beats the LPU and gets close to Cerebras
Lossless Inference
The fastest way to serve DeepSeek V4 Flash
$0.09 and $290.12: What Actually Moves Your Inference Bill
Serving Kimi K3 on vLLM was hard. Here is what we measured.
vLLM vs SGLang vs TensorRT-LLM: a reproducible benchmark
Deploy your first optimized model, measured before you ship
Describe the goal. RunInfra builds and optimizes the stack.
End-to-end encryption
Isolated GPU infrastructure
No training on your data
SOC 2 Type II