Bug 2517595 (CVE-2026-71486)

Summary:	CVE-2026-71486 vllm: vLLM: Denial of Service via unbounded token ID decoding
Product:	[Other] Security Response	Reporter:	OSIDB Bzimport <bzimport>
Component:	vulnerability	Assignee:	Product Security <prodsec-ir-bot>
Status:	NEW ---	QA Contact:
Severity:	medium	Docs Contact:
Priority:	medium
Version:	unspecified	CC:	alinfoot, bbrownin, dtrifiro, weaton
Target Milestone:	---	Keywords:	Security
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	---
Doc Text:	A flaw was found in vLLM. An authenticated API client can exploit derender endpoints by providing oversized GenerateResponse objects. These endpoints process the objects without enforcing output bounds or response-size limits. This can lead to excessive consumption of CPU and memory resources, resulting in a Denial of Service (DoS).	Story Points:	---
Clone Of:		Environment:
Last Closed:		Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description OSIDB Bzimport 2026-08-17 20:26:27 UTC

vLLM is an inference and serving engine for large language models. Prior to 0.26.0, the /v1/completions/derender and /v1/chat/completions/derender endpoints accept caller-supplied GenerateResponse objects whose generate_responses, choices, token_ids, prompt_logprobs, logprobs.content, top_logprobs, and routed_experts structures are processed by OnlineDerenderer and tokenizer.decode before max_model_len, max_tokens, max_num_seqs, or response-size limits are enforced, allowing an authenticated API client to consume excessive CPU and memory and produce oversized responses. This issue is fixed in version 0.26.0.