Pull requests

• [Klaud Cold] Update GLM-5.2 FP4 B200 Dynamo-SGLang AgentX agg image to v0.5.22-cu130 / [Klaud Cold] 将 GLM-5.2 FP4 B200 Dynamo-SGLang AgentX 聚合镜像更新至 v0.5.22-cu130

  #4370·Klaud-Cold opened Oct 21, 2026Collaborator·

  ·
  2

• [Klaud Cold] Update Kimi-K3 FP4 B200 Dynamo-vLLM AgentX DSpark image to v0.32.0 / [Klaud Cold] 将 Kimi-K3 FP4 B200 Dynamo-vLLM AgentX DSpark 镜像更新至 v0.32.0

  #4369·Klaud-Cold opened Oct 21, 2026Collaborator·

  ·
  2

• [Klaud Cold] Update DSR1 FP8 MI325X SGLang MTP image to v0.5.22-rocm720-mi30x / [Klaud Cold] 将 DSR1 FP8 MI325X SGLang MTP 镜像更新至 v0.5.22-rocm720-mi30x

  full-sweep-fail-fast
  #4368·Klaud-Cold opened Oct 21, 2026Collaborator·

  ·
  8

• [AMD][MI35X] 1021 DSV41 sglang mtp agentic

  agentxAMDnon-canary-full-sweep-enabled
  #4367·1am9trash opened Oct 21, 2026Collaborator·

  ·
  3

• [AgentX] Enable unified memory for B200 Muse Spark / [AgentX] 为 B200 Muse Spark 启用 unified memory

  full-sweep-fail-fast
  #4366·Fridge003 opened Oct 21, 2026Collaborator·6 of 11 tasks·

  ·
  2

• [Klaud Cold] Update DeepSeek-V4.1-Flash FP4 B200 SGLang AgentX image to nightly-dev-cu13-20261021-c1c287a4 / [Klaud Cold] 将 DeepSeek-V4.1-Flash FP4 B200 SGLang AgentX 镜像更新至 nightly-dev-cu13-20261021-c1c287a4

  full-sweep-fail-fast
  #4365·Klaud-Cold opened Oct 21, 2026Collaborator·

  ·
  8

• [Klaud Cold] Update DSR1 FP8 MI300X SGLang image to v0.5.22-rocm720-mi30x / [Klaud Cold] 将 DSR1 FP8 MI300X SGLang 镜像更新至 v0.5.22-rocm720-mi30x

  full-sweep-fail-fast
  #4363·Klaud-Cold opened Oct 21, 2026Collaborator·

  ·
  8

• [NVIDIA][AgentX] Kimi-K3 B300 TP8 sweep with lazy DRAM KV offload / [NVIDIA][AgentX] Kimi-K3 B300 TP8 惰性 DRAM KV 卸载扫描

  full-sweep-fail-fast
  #4362·Oseltamivir opened Oct 21, 2026Collaborator·

  ·
  4

• [NVIDIA][AgentX] Kimi-K3 B300 TP8 sweep with a GPU-only KV cache / [NVIDIA][AgentX] Kimi-K3 B300 TP8 仅使用 GPU KV 缓存的扫描

  full-sweep-fail-fast
  #4361·Oseltamivir opened Oct 21, 2026Collaborator·

  ·
  6

• [Klaud Cold] Update DSR1 FP8 H200 TRT-LLM image to 1.4.0rc1 / [Klaud Cold] 将 DSR1 FP8 H200 TRT-LLM 镜像更新至 1.4.0rc1

  full-sweep-fail-fast
  #4359·Klaud-Cold opened Oct 21, 2026Collaborator·

  ·
  9

• [Klaud Cold] Update DSR1 FP4 MI355X ATOM image to v0.1.8 / [Klaud Cold] 将 DSR1 FP4 MI355X ATOM 镜像更新至 v0.1.8

  full-sweep-fail-fast
  #4358·Klaud-Cold opened Oct 21, 2026Collaborator·

  ·
  8

• MiniMax-M3 NVFP4 B300 fp4 vLLM AgentX: tuned config-only curve on nightly-ac68c308 / MiniMax-M3 NVFP4 B300 fp4 vLLM AgentX：基于 nightly-ac68c308 的纯配置调优曲线

  full-sweep-enabled
  #4357·hshrivastava-droid opened Oct 20, 2026Collaborator·5 of 10 tasks·

  ·
  1

• perf(qwen3.5): move GB300 TRT AgentX to TRT-LLM 1.3.0a0.dev202610200641 / perf(qwen3.5): GB300 TRT AgentX 切换到 TRT-LLM 1.3.0a0.dev202610200641

  #4356·richardhuo-nv opened Oct 20, 2026Collaborator·

  ·
  14

• [AgentX] Add GB300 DeepSeek-V4.1-Flash disagg 10P1D c8400 point

  #4352·cquil11 opened Oct 20, 2026Collaborator·

  ·
  3

• [AgentX] Retune Muse Spark Open B200 SGLang per concurrency / [AgentX] 按并发调优 Muse Spark Open B200 SGLang

  full-sweep-fail-fast
  #4350·shyeh25 opened Oct 20, 2026Collaborator·4 of 11 tasks·

  ·
  8

• [NVIDIA][AgentX] Kimi-K3 B300 TP8 sweep conc 1-64 on a 1.5 TB DRAM tier / [NVIDIA][AgentX] Kimi-K3 B300 TP8 在 1.5 TB DRAM 层上扫描 conc 1-64

  full-sweep-fail-fast
  #4349·Oseltamivir opened Oct 20, 2026Collaborator·

  ·
  5

• [NVIDIA][AgentX] Kimi-K3 B300 TP8 sweep conc 1-64 on the full 3 TB DRAM tier / [NVIDIA][AgentX] Kimi-K3 B300 TP8 在完整 3 TB DRAM 层上扫描 conc 1-64

  full-sweep-fail-fast
  #4348·Oseltamivir opened Oct 20, 2026Collaborator·

  ·
  6

• [Klaud Cold] Update Qwen3.5 FP8 MI355X ATOM image to v0.1.8 / [Klaud Cold] 将 Qwen3.5 FP8 MI355X ATOM 镜像更新至 v0.1.8

  full-sweep-fail-fast
  #4347·Klaud-Cold opened Oct 20, 2026Collaborator·

  ·
  11

• Add B300 Dynamo+SGLang AgentX configs for DeepSeek V4 with Mooncake EFA KV transfer / 添加使用 Mooncake EFA KV 传输的 DeepSeek V4 B300 Dynamo+SGLang AgentX 配置

  full-sweep-fail-fast
  #4346·nvpohanh opened Oct 20, 2026Collaborator·7 of 11 tasks·

  ·
  7

• config(musespark): add GB300 Dynamo SGLang AgentX aggregated points / config(musespark): 新增 GB300 Dynamo SGLang AgentX 聚合配置点

  full-sweep-fail-fast
  #4344·shyeh25 opened Oct 20, 2026Collaborator·4 of 11 tasks·

  ·
  10

• config(musespark): tune GB200 SGLang AgentX low-concurrency settings / config(musespark): 调整 GB200 SGLang AgentX 低并发参数

  full-sweep-fail-fast
  #4343·shyeh25 opened Oct 20, 2026Collaborator·4 of 11 tasks·

  ·
  8

• config(musespark): tune GB300 SGLang AgentX settings per concurrency / config(musespark): 按并发调整 GB300 SGLang AgentX 参数

  full-sweep-fail-fast
  #4342·shyeh25 opened Oct 20, 2026Collaborator·4 of 11 tasks·

  ·
  8

• config(musespark): tune B300 SGLang AgentX settings per concurrency / config(musespark): 按并发调整 B300 SGLang AgentX 参数

  full-sweep-fail-fast
  #4341·shyeh25 opened Oct 20, 2026Collaborator·4 of 11 tasks·

  ·
  8

• [Klaud Cold] Update DeepSeek-V4-Pro FP4 GB300 Dynamo-SGLang AgentX disagg image to v0.5.22-cu130 / [Klaud Cold] 将 DeepSeek-V4-Pro FP4 GB300 Dynamo-SGLang AgentX 分离式镜像更新至 v0.5.22-cu130

  #4340·Klaud-Cold opened Oct 20, 2026Collaborator·

  ·
  14

• feat(agentx): run AgentX on upstream AIPerf / feat(agentx): 在上游 AIPerf 上运行 AgentX

  #4338·FrankD412 opened Oct 20, 2026Collaborator·7 of 11 tasks·

  ·