Servers and applications

CVE-2026-126398

vLLM 0.28.0 before 0.30.0 Denial of Service via Prompt Embeds on XD-RoPE Models

Severity: High 8.2
Published: 13 Nov 2026
Last changed: 13 Nov 2026
Added to UpHeron: 13 Nov 2026
Fixed in: 0.30.0
Weakness: CWE-617
Attack: Over the network · Only under certain conditions · No login needed · No user action needed
Attacker can: Take it down
CVSS vector: CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N
Sources: CVE Program

Description

vLLM from 0.28.0 before 0.30.0 contains a reachable assertion vulnerability in GPUModelRunner._init_xdrope_positions that allows remote attackers to crash the engine by submitting embeddings-only requests. Attackers can send a /v1/completions request containing only prompt_embeds to XD-RoPE models with --enable-prompt-embeds, terminating the EngineCore and API server for all clients.

From the CVE Program.

Affected software

Software	Affected versions
vLLMServers and applications · `cna:vllm-project\|vllm`	From 0.28.0 before 0.30.0

References

CVE.org https://www.cve.org/CVERecord?id=CVE-2026-126398