From 1a37e57fb1ae9b937db566df0f42d6a87a83a2ac Mon Sep 17 00:00:00 2001 From: zijiexia <37504505+zijiexia@users.noreply.github.com> Date: Fri, 24 Apr 2026 00:06:30 -0700 Subject: [PATCH] [codex] docs: note H200 DeepSeek-V4 checkpoint (#23628) --- docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-V4.mdx | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-V4.mdx b/docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-V4.mdx index 706b9d251..80114ad0b 100644 --- a/docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-V4.mdx +++ b/docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-V4.mdx @@ -99,6 +99,10 @@ Please refer to the [official SGLang installation guide](../../../docs/get-start SGLang supports three main serving recipes for DeepSeek-V4 with different latency/throughput trade-offs (`low-latency`, `balanced`, `max-throughput`), plus specialized recipes for long-context (`cp`, prefill context-parallel) and prefill/decode disaggregation (`pd-disagg`). The interactive generator below emits the exact launch command for any `(hardware, variant, recipe)` combination. + +For H200 GPU deployments, use the SGLang checkpoint under `sgl-project`, not the default DeepSeek checkpoint. + + ### 3.1 Basic Configuration **Interactive Command Generator**: Use the selector below to generate the deployment command for your hardware + recipe combination.