Gimlet Labs

First multi-silicon inference cloud provider optimizing LLM workloads across GPUs, CPUs, and specialized architectures. Its software splits inference stages to maximize throughput and reduce latency for agentic applications, achieving up to tenfold performance gains at equivalent power consumption.