[Klaud Cold] qwen3.8next-fp8-h200-sglang-agentic-mtp: Qwen3.8-Flash-Next FP8 SGLang AgentX on H200 / H200 上 Qwen3.8-Flash-Next FP8 SGLang AgentX 配方 - #2753

Merged

functionstackx merged 8 commits into

mainfrom

feat/qwen3.8next-fp8-h200-sglang-agentic-mtp

Aug 27, 2026

Commits on Aug 27, 2026

qwen3.8next-fp8-h200-sglang-agentic-mtp: day-zero Qwen3.8-Flash-Next AgentX on H200

functionstackx
and
claude
committed
9c88dd9
Fill perf-changelog pr-link for #2753 / 补全 #2753 的 perf-changelog pr-link

functionstackx
and
claude
committed
e3a111c
Use the cookbook's verified TP4/EP4 command and the committed golden AL

functionstackx
and
claude
committed
65a1a76
Define MAX_RUNNING_REQUESTS and CUDA_GRAPH_MAX_BS on the H200 arm

functionstackx
and
claude
committed
1fa09be
Use a float32 Mamba SSM state so the NEXTN verify kernel accepts it

functionstackx
and
claude
committed
f1da353
Separate the appended changelog entry from history with a blank line

functionstackx
and
claude
committed
d08ce6c
chore: refresh PR #2753 for sweep reuse [skip-sweep]
functionstackx
committed
c1e509d
chore: refresh PR #2753 for sweep reuse [skip-sweep]
functionstackx
committed
2eccb1a