---
cited_by:
  - '[[thoughts/inference/kimi-k3-prs#4-pr-reading-order|kimik3 §4]]'
date: '2026-11-11'
kind: pr
local:
  - file:///Users/aarnphm/workspace/vllm/vllm/tool_parsers/structural_tag_registry.py
  - file:///Users/aarnphm/workspace/vllm/vllm/config/model.py
  - file:///Users/aarnphm/workspace/vllm/vllm/parser/parser_manager.py
  - file:///Users/aarnphm/workspace/vllm/vllm/entrypoints/chat_utils.py
merged: '2026-07-29'
number: 50093
source:
  - https://github.com/vllm-project/vllm/blob/f5a7cce9b6a61f4d995629a7418c7ea822e34a64/vllm/tool_parsers/kimi_k3_tool_parser.py
  - https://github.com/vllm-project/vllm/blob/f5a7cce9b6a61f4d995629a7418c7ea822e34a64/vllm/reasoning/kimi_k3_reasoning_parser.py
  - https://github.com/vllm-project/vllm/blob/f5a7cce9b6a61f4d995629a7418c7ea822e34a64/vllm/tool_parsers/structural_tag_registry.py
  - https://github.com/vllm-project/vllm/blob/f5a7cce9b6a61f4d995629a7418c7ea822e34a64/vllm/renderers/kimi_k3.py
  - https://github.com/vllm-project/vllm/blob/f5a7cce9b6a61f4d995629a7418c7ea822e34a64/vllm/parser/kimi_k3.py
  - https://github.com/vllm-project/vllm/blob/f5a7cce9b6a61f4d995629a7418c7ea822e34a64/vllm/config/model.py
  - https://github.com/vllm-project/vllm/blob/f5a7cce9b6a61f4d995629a7418c7ea822e34a64/vllm/parser/parser_manager.py
  - https://github.com/vllm-project/vllm/blob/f5a7cce9b6a61f4d995629a7418c7ea822e34a64/vllm/entrypoints/chat_utils.py
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#50093 [Model] Add Kimi K3 support: Python frontend [2/2]'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/50093
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/50093
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/50093.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
Extract the Kimi K3 Python frontend from [#50000](https://github.com/aarnphm/aarnphm.github.io/issues/50000) as the top layer of the independent frontend stack rooted at `main` through [#50104](https://github.com/aarnphm/aarnphm.github.io/issues/50104). This layer adds Python XTML rendering, reasoning and tool parsing, structural-tag handling, and streaming… · full dump: [[thoughts/inference/vllm-prs/50093|#50093]]

