Show HN: Flide: Make Jev-like decision models up to 9.5x faster

Flide is a high-performance inference engine for Jev-like decision models, achieving up to a 9.5x speedup over vLLM and SGLang. It supports 30+ popular local Jev-like decision models and provides exact state forking, ensuring every answer is the model's original output. Flide serves the Jev /v1/systemone API and is available for installation via pip. It is released under the Apache-2.0 license and uses a kernel derived from flash-linear-attention.