{"id":"CVE-2026-100647","title":"vLLM versions before 0.29.0 contain a denial-of-service vulnerability in the cache_salt parameter accepted on OpenAI-compatible and Anthropic API endpoints, which lacks maximum length validation and is processed on the single EngineCore …","summary":"vLLM versions before 0.29.0 contain a denial-of-service vulnerability in the cache_salt parameter accepted on OpenAI-compatible and Anthropic API endpoints, which lacks maximum length validation and is processed on the single EngineCore …","severity":"medium","cvss":5.3,"cvssVector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L","cwe":["CWE-20","CWE-1322"],"vendor":"vllm","product":"vllm","affected":["vllm < 0.29.0"],"patched":["vllm 0.29.0"],"published":"2026-09-26","updated":"2026-10-06","sourceUpdated":"2026-10-06T19:46:06.527","source":"NVD","sourceUrl":"https://nvd.nist.gov/vuln/detail/CVE-2026-100647","references":[{"url":"https://github.com/vllm-project/vllm/security/advisories/GHSA-wpww-v874-ph2p","label":"disclosure@vulncheck.com"},{"url":"https://www.vulncheck.com/advisories/vllm-before-0.29.0-cpu-exhaustion-via-unbounded-cache-salt","label":"disclosure@vulncheck.com"},{"url":"https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-100647.json"},{"url":"https://access.redhat.com/security/cve/CVE-2026-100647"},{"url":"https://bugzilla.redhat.com/show_bug.cgi?id=2541900"},{"url":"https://www.cve.org/CVERecord?id=CVE-2026-100647"},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-100647"}],"tags":["nvd","cve.org","csaf","vex","red-hat"],"epss":0.00348,"epssPercentile":0.28272,"ssvc":{"exploitation":"none","automatable":"yes","technicalImpact":"partial","timestamp":"2026-09-28T18:26:37.598538Z"},"ingestedAt":"2026-09-26T13:33:20.921Z","slug":"CVE-2026-100647","body":"## Overview\n\nvLLM versions before 0.29.0 contain a denial-of-service vulnerability in the cache_salt parameter accepted on OpenAI-compatible and Anthropic API endpoints, which lacks maximum length validation and is processed on the single EngineCore scheduler thread. Unauthenticated attackers can send HTTP requests with multi-hundred-megabyte salt values that trigger expensive pickle serialization and SHA-256 hashing, stalling the scheduler thread and denying service to all concurrent requests.\n\n## Affected\n\n- `vllm < 0.29.0`\n\n## Remediation\n\nUpgrade past the affected range:\n\n- `vllm 0.29.0`\n\n## Vendor advisories\n\n- **Red Hat VEX** · Moderate · affected: Red Hat AI Inference Server, Red Hat Enterprise Linux AI (RHEL AI) 3, Red Hat OpenShift AI (RHOAI) · no fix planned: Red Hat AI Inference Server, Red Hat Enterprise Linux AI (RHEL AI) 3, Red Hat OpenShift AI (RHOAI) · updated 2026-10-06 · [vex](https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-100647.json)","depth":"sunlit","depthScore":29,"depthScoreParts":{"impact":29.2,"likelihood":0.1,"exploitation":0,"ransomware":0},"changes":[]}