kekzl/imp

kekzl/imp 是一款面向单张 NVIDIA Blackwell GPU 的 LLM 服务器，提供 API、模型输入格式以及公开的性能测量数据。

GitHub 项目说明: LLM server for NVIDIA Blackwell only (GeForce RTX 50, RTX PRO). On one RTX 5090, Qwen3.8-27B one user: 96.3 tok/s, 1.38x vLLM without speculative decoding, 1.23x llama.cpp on the same GGUF. Serves the 123.6 GiB Qwen3.8-Flash-Next from a 32 GB card at 92.1 tok/s. OpenAI and Anthropic API, tool calls, one docker run.