Inferact, built by vLLM’s creators, makes AI inference cheaper and faster by expanding open-source support across 500+ model architectures and 200+ accelerator types.