AION
Open-source releaseEfficiency & Inference · Large Language Models · Training & Scaling1 source · Jan 23, 2026

sgl-project/sglang v0.5.8

Qwen3-VL-Embedding & Qwen3-VL-Reranker model support: #16635, #16403

OpenAINVIDIADeepSeekQwenDeepSeek modelsHugging Face
← Back to brief

Key points

  • Day 0 Support for GLM 4.7 Flash: #17247
  • Context Parallelism Optimization with support for fused MoE, multi-batch, and FP8 KV cache: #13959
  • Added dependencies for tvm-ffi and quack-kernels: #17075
  • Mooncake transfer engine updated to 0.3.8.post1: #16792

Sources (1)

  • [1]sgl-project/sglang v0.5.8
    GitHub: sgl-project/sglang · Jan 23, 10:09 PM
    * Qwen3-VL-Embedding & Qwen3-VL-Reranker model support: #16635, #16403
    * Day 0 Support for GLM 4.7 Flash: #17247

Extractive summary: sentences quoted from the sources.