---
cited_by:
  - '[[thoughts/inference/deepseek-v4.1-serving#111-vl|dsv41 §1.11]]'
date: '2026-11-11'
kind: pr
local: []
merged: '2026-09-20'
number: 56625
source:
  - https://github.com/vllm-project/vllm/blob/27757dde020ecda4f9b0e2c5ca2df1c29badc82f/vllm/models/deepseek_v41/common/vl_cudagraph.py
  - https://github.com/vllm-project/vllm/blob/27757dde020ecda4f9b0e2c5ca2df1c29badc82f/vllm/models/deepseek_v4/common/vision.py
  - https://github.com/vllm-project/vllm/blob/27757dde020ecda4f9b0e2c5ca2df1c29badc82f/vllm/models/deepseek_v41/amd/vl_model.py
  - https://github.com/vllm-project/vllm/blob/27757dde020ecda4f9b0e2c5ca2df1c29badc82f/vllm/models/deepseek_v41/nvidia/vl_model.py
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#56625 [DSV4.1] Add encoder cuda graph support for deepseek-v4.1-flash'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/56625
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/56625
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/56625.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
Enable CUDA graph capture/replay for the DeepSeek-V4.1 vision encoder (ViT + · full dump: [[thoughts/inference/vllm-prs/56625|#56625]]

