Claim listing
kekzl/imp
LLM server for one RTX 5090. Runs Qwen3.8-Flash-Next (512-expert MoE) on a single 32 GB card at 75-80 tok/s. Decodes 17-191 % faster than llama.cpp on the same models. Built for agents: tool calls, long context, many streams at once.
Claim your listing to add a tagline, logo, and category. Verified maintainers get a Verified Publisher badge and priority placement on the AgentRank index.
Leave your email to claim this listing. GitHub verification coming soon.