{"id":"CVE-2026-100652","aliases":["EUVD-2026-87746","GHSA-gxcm-7c4h-f56c","GHSA-qff2-492f-9fm4","PYSEC-2026-4185"],"status":"published","title":"vLLM 0.22.0 through 0.23.0 Denial of Service via stop_token_ids","description":"vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with min_tokens greater than zero and out-of-vocabulary stop_token_ids to trigger CUDA tensor indexing failures that leave EngineCore in a fatal state requiring service restart.","severity":{"version":"4.0","score":8.2,"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","level":"high","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-09","score":0.00419,"percentile":0.36529,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":true},"cwes":["CWE-20"],"published":"2026-09-26T13:23:21Z","modified":"2026-10-28T02:31:09Z","fixes":[{"ecosystem":"PyPI","package":"vllm","fixed_in":["0.24.0"],"sources":["osv"]}],"affected":[{"source":"osv","ecosystem":"PyPI","package":"vllm","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0.22.0"},{"fixed":"0.24.0"}]}],"versions":["0.22.0","0.22.1","0.23.0"]}],"products":[{"vendor":"vllm","product":"vllm"},{"vendor":"vllm-project","product":"vllm"}],"references":[{"url":"https://github.com/vllm-project/vllm/security/advisories/GHSA-qff2-492f-9fm4","tags":["vendor-advisory","Exploit","Mitigation","Vendor Advisory","advisory","evidence"]},{"url":"https://www.vulncheck.com/advisories/vllm-0.22.0-through-0.23.0-denial-of-service-via-stop-token-ids","tags":["third-party-advisory","Third Party Advisory","advisory"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-100652","tags":["advisory"]},{"url":"https://github.com/advisories/GHSA-gxcm-7c4h-f56c"},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/100xxx/CVE-2026-100652.json","tags":["advisory"]}],"sources":{"cve":[{"id":"CVE-2026-100652","title":"vLLM 0.22.0 through 0.23.0 Denial of Service via stop_token_ids","description":"vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with min_tokens greater than zero and out-of-vocabulary stop_token_ids to trigger CUDA tensor indexing failures that leave EngineCore in a fatal state requiring service restart.","status":"published","published":"2026-09-26T13:23:21.865Z","modified":"2026-09-28T16:42:13.668Z","cvss":[{"version":"4.0","score":8.2,"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","level":"high","assessor":"cna"},{"version":"3.1","score":5.9,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:H","level":"medium","assessor":"cna"}],"cwes":["CWE-20"],"updated_at":"2026-11-02T14:20:06.925367Z"}],"euvd":[{"id":"EUVD-2026-87746","description":"vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with min_tokens greater than zero and out-of-vocabulary stop_token_ids to trigger CUDA tensor indexing failures that leave EngineCore in a fatal state requiring service restart.","status":"published","published":"2026-09-26T13:23:21Z","modified":"2026-09-28T16:42:13Z","cvss":[{"version":"4.0","score":8.2,"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","level":"high"}],"updated_at":"2026-11-02T17:27:53.091489Z"}],"ghsa":[{"id":"GHSA-gxcm-7c4h-f56c","title":"vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in...","description":"vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with min_tokens greater than zero and out-of-vocabulary stop_token_ids to trigger CUDA tensor indexing failures that leave EngineCore in a fatal state requiring service restart.","status":"published","published":"2026-09-26T15:31:18Z","modified":"2026-09-26T15:31:22Z","cvss":[{"version":"3.1","score":5.9,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:H","level":"medium"},{"version":"4.0","score":8.2,"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/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X","level":"high"}],"cwes":["CWE-20"],"updated_at":"2026-11-02T15:12:30.047516Z"}],"nvd":[{"id":"CVE-2026-100652","description":"vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with min_tokens greater than zero and out-of-vocabulary stop_token_ids to trigger CUDA tensor indexing failures that leave EngineCore in a fatal state requiring service restart.","status":"published","published":"2026-09-26T14:16:47.81Z","modified":"2026-10-06T19:22:01.53Z","cvss":[{"version":"4.0","score":8.2,"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/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X","level":"high","assessor":"secondary"},{"version":"3.1","score":5.9,"vector":"CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:H","level":"medium","assessor":"secondary"},{"version":"3.1","score":7.5,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H","level":"high","assessor":"primary"}],"cwes":["CWE-20"],"updated_at":"2026-11-02T14:12:52.081323Z"}],"osv":[{"id":"CVE-2026-100652","title":"vLLM 0.22.0 through 0.23.0 Denial of Service via stop_token_ids","description":"vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with min_tokens greater than zero and out-of-vocabulary stop_token_ids to trigger CUDA tensor indexing failures that leave EngineCore in a fatal state requiring service restart.","status":"published","published":"2026-09-26T13:23:21.865Z","modified":"2026-10-28T02:31:09.67017831Z","cvss":[{"version":"4.0","score":8.2,"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","level":"high"}],"cwes":["CWE-20"],"updated_at":"2026-11-02T15:44:53.490346Z"},{"id":"PYSEC-2026-4185","description":"vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with min_tokens greater than zero and out-of-vocabulary stop_token_ids to trigger CUDA tensor indexing failures that leave EngineCore in a fatal state requiring service restart.","status":"published","published":"2026-09-26T14:16:47.81Z","modified":"2026-10-07T10:00:03.378975746Z","cvss":[{"version":"3.1","score":7.5,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H","level":"high"}],"updated_at":"2026-11-02T15:52:57.671918Z"}]}}