AION
Open-source releaseEfficiency & Inference · Large Language Models · Image, Video & 3D Generation1 source · Jun 13, 2026

sgl-project/sglang v0.5.13

DeepSeek V4 — context parallelism & sparse-attention kernels: Building on the v0.5.12 Day-0 path, v0.5.13 extends DeepSeek-V4 to context-parallel serving and adds its sparse-attention kernels:

OpenAINVIDIADeepSeekQwenDeepSeek modelsAMD
← Back to brief

Sources (1)

  • [1]sgl-project/sglang v0.5.13
    GitHub: sgl-project/sglang · Jun 13, 12:17 AM
    **DeepSeek V4 — context parallelism & sparse-attention kernels**: Building on the v0.5.12 Day-0 path, v0.5.13 extends DeepSeek-V4 to context-parallel serving and adds its sparse-attention kernels:

Extractive summary: sentences quoted from the sources.