CVE-2026-44222High· 7.5▾ TwilightA flaw was found in vLLM, an inference and serving engine for large language models (LLMs). This vulnerability allows unauthenticated attackers to cause a Denial of Service (DoS) by supplying image or video placeholder sequences without ma…
▾ Twilight zone — High severity, or a signal on a lesser flaw
impact 41.3 · likelihood 0.1 · exploitation 0
Need a working PoC? Pro members can cast a request and our team develops one — it lands right here.
Exploit-prediction probability, daily snapshots since Aug 3.
Disclosure to exploitation, from the record and what we observed since indexing it.
Disclosed via CSAF
0.4%
0.4% → 0.5%
Last analysed / modified upstream
6.5 → 7.5
medium → high
A flaw was found in vLLM, an inference and serving engine for large language models (LLMs). This vulnerability allows unauthenticated attackers to cause a Denial of Service (DoS) by supplying image or video placeholder sequences without matching data, leading to an unhandled error and worker termination. Additionally, text-only prompts containing special tokens can be misinterpreted as control commands, potentially leading to unexpected behavior.
vllm: vLLM: Denial of Service via malformed multimodal input or token injection — rated Important by Red Hat. Released 2026-05-12, updated 2026-10-02.
Affected:
Fixed:
No fix planned:
Not affected:
For more information visit https://access.redhat.com/errata/RHSA-2026:61627 https://access.redhat.com/errata/RHSA-2026:61627 For more information visit https://access.redhat.com/errata/RHSA-2026:61629 https://access.redhat.com/errata/RHSA-2026:61629 For more information visit https://access.redhat.com/errata/RHSA-2026:60363 https://access.redhat.com/errata/RHSA-2026:60363
Workarounds / mitigations:
Affected packages:
vllm >= 0.6.1, < 0.20.0Patched in:
vllm 0.20.0Field changes observed since this record was first indexed.
Connected by shared product, vendor, weakness, or advisory.
CVE-2026-55380High· 7.5python-pillow: Pillow: Denial of Service via crafted GD 2.x image file (CVE-2026-55380)
CVE-2026-59936Medium· 6.5pypdf: pypdf: Denial of Service via crafted PDF inline image (CVE-2026-59936)
CVE-2026-59935Medium· 6.5pypdf: pypdf: Denial of Service via crafted PDF with unterminated inline image (CVE-2026-59935)
CVE-2026-59938Medium· 5.3pypdf: pypdf: Possible large memory usage for wrong image dimensions (CVE-2026-59938)
CVE-2026-59937Medium· 6.5pypdf: pypdf: Denial of Service via crafted PDF with malformed cross-reference streams (CVE-2026-59937)
CVE-2026-6855High· 7.1instructlab: InstructLab: Path traversal allows arbitrary directory creation and file write (CVE-2026-6855)