llmux 0.7.7

Zero-reload model switching for vLLM - manages multiple models on shared GPU
Documentation