PT-2025-41232 · Vllm · Vllm

CVE-2025-61620 · Published 2025-10-07 · Updated 2026-07-07

Positive Technologies (AI Summary)

CVSS v3.1

6.5

Medium

Base vector

AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H

Name of the Vulnerable Software and Affected Versions vLLM (affected versions not specified)

Description A resource-exhaustion denial-of-service issue exists in multiple endpoints of the OpenAI-Compatible Server. The server allows users to specify Jinja templates through the chat template and chat template kwargs parameters. Because Jinja templates support complex constructs like loops and nested iterations, a crafted template can consume excessive CPU and memory resources, leading to a service outage.

Technical exploitation occurs because the server uses dict.update to merge user-supplied chat template kwargs into the arguments for the apply hf chat template() function. This allows an attacker to override the chat template key even if the primary chat template parameter is restricted.

Recommendations At the moment, there is no information about a newer version that contains a fix for this vulnerability.

Allocation of Resources Without Limits

RCE

Resource Exhaustion

Weakness Enumeration

CWE-770

CWE-20

CWE-400

Related Identifiers

CVE-2025-61620

GHSA-6FVQ-23CW-5628

PYSEC-2026-2013

Affected Products

Vllm

References · 9

• https://osv.dev/vulnerability/PYSEC-2026-2013 · Vendor Advisory
• https://nvd.nist.gov/vuln/detail/CVE-2025-61620 · Security Note
• https://osv.dev/vulnerability/GHSA-6fvq-23cw-5628 · Vendor Advisory
• https://github.com/vllm-project/vllm · Note
• https://github.com/vllm-project/vllm/pull/25794 · Note
• https://github.com/vllm-project/vllm/commit/7977e5027c2250a4abc1f474c5619c40b4e5682f · Note
• https://github.com/vllm-project/vllm/security/advisories/GHSA-6fvq-23cw-5628 · Note
• https://github.com/advisories/GHSA-6fvq-23cw-5628 · Note
• https://pypi.org/project/vllm · Note