Medium severity6.5OSV Advisory· Published Jul 6, 2026· Updated Jul 7, 2026
CVE-2026-55514
CVE-2026-55514
Description
vLLM is a library for LLM inference and serving. From 0.12.0 to before 0.24.0, sending a pure prompt embeds payload in a /v1/completions request with a model using M-RoPE causes EngineCore to fail an assertion and fatally crash, shutting down the entire server application. Any remote user who is authorized to make a /v1/completions request can make such a request and induce a crash. This issue is fixed in version 0.24.0.
Affected packages
Versions sourced from the GitHub Security Advisory.
| Package | Affected versions | Patched versions |
|---|---|---|
vllmPyPI | >= 0.12.0, < 0.24.0 | 0.24.0 |
Affected products
3Patches
Vulnerability mechanics
References
7- github.com/vllm-project/vllm/commit/470229c37efaf69c86e8bc97482b0b1ff7551c65nvdPatchWEB
- github.com/vllm-project/vllm/pull/45252nvdIssue TrackingPatchWEB
- github.com/advisories/GHSA-33cg-gxv8-3p8gghsaADVISORY
- github.com/vllm-project/vllm/security/advisories/GHSA-33cg-gxv8-3p8gnvdVendor AdvisoryWEB
- nvd.nist.gov/vuln/detail/CVE-2026-55514ghsaADVISORY
- github.com/pypa/advisory-database/tree/main/vulns/vllm/PYSEC-2026-2303.yamlghsaWEB
- github.com/vllm-project/vllm/releases/tag/v0.24.0nvdRelease NotesWEB
News mentions
1- VLLM: Four DoS Vulnerabilities Disclosed Together on July 6, 2026Vypr Intelligence · Jul 6, 2026