← ArchiveGitHub
data as of 2026-10-29
Off board · last on board 9d agoPython · Apache-2.0
syv-ai/HyperQwen★ 2.4k

Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.

stars
2.4k
commits / 13w
571
Contributors
4
Star growth · 13d+387
  • fork 309
  • open issues 9
hubtrend.app/i/github/syv-ai/HyperQwenpushed 17d ago · fetched 10d ago

Repository panorama

from public APIs, independent of our rankings

Health

activity, response time, maintainers
Last push
17d ago
Issue response
3.6d

median over 100 closed

Open issues
9
Apache-2.0Pythonkv-cachellm-inferencelocal-llmquantizationqwenqwen3rtx-3090speculative-decoding
Open on GitHub ↗
syv-ai/HyperQwen · 2.4k stars · HubTrend