Moderate severityNVD Advisory· Published Nov 21, 2025· Updated Nov 24, 2025
vLLM vulnerable to DoS via large Chat Completion or Tokenization requests with specially crafted `chat_template_kwargs`
CVE-2025-62426
Description
vLLM is an inference and serving engine for large language models (LLMs). From version 0.5.5 to before 0.11.1, the /v1/chat/completions and /tokenize endpoints allow a chat_template_kwargs request parameter that is used in the code before it is properly validated against the chat template. With the right chat_template_kwargs parameters, it is possible to block processing of the API server for long periods of time, delaying all other requests. This issue has been patched in version 0.11.1.
AI Insight
LLM-synthesized narrative grounded in this CVE's description and references.
Affected packages
Versions sourced from the GitHub Security Advisory.
| Package | Affected versions | Patched versions |
|---|---|---|
vllmPyPI | >= 0.5.5, < 0.11.1 | 0.11.1 |
Affected products
3- osv-coords2 versions
< 25.9.0_git20251112-r1+ 1 more
- (no CPE)range: < 25.9.0_git20251112-r1
- (no CPE)range: >= 0.5.5, < 0.11.1
Patches
Vulnerability mechanics
References
7- github.com/advisories/GHSA-69j4-grxj-j64pghsaADVISORY
- nvd.nist.gov/vuln/detail/CVE-2025-62426ghsaADVISORY
- github.com/vllm-project/vllm/blob/2a6dc67eb520ddb9c4138d8b35ed6fe6226997fb/vllm/entrypoints/chat_utils.pyghsax_refsource_MISCWEB
- github.com/vllm-project/vllm/blob/2a6dc67eb520ddb9c4138d8b35ed6fe6226997fb/vllm/entrypoints/openai/serving_engine.pyghsax_refsource_MISCWEB
- github.com/vllm-project/vllm/commit/3ada34f9cb4d1af763fdfa3b481862a93eb6bd2bghsax_refsource_MISCWEB
- github.com/vllm-project/vllm/pull/27205ghsax_refsource_MISCWEB
- github.com/vllm-project/vllm/security/advisories/GHSA-69j4-grxj-j64pghsax_refsource_CONFIRMWEB
News mentions
0No linked articles in our index yet.