{"id":"CVE-2026-54234","summary":"vLLM: Remote DoS in vLLM via Invalid Recovered Token Reinjection","details":"vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, a frontend-legal multi-request speculative decoding workload can cause the rejection sampler to produce a recovered token equal to the model vocabulary size boundary value, which is then converted to negative one when the engine selects the next live token for a request and is written back into the drafter's input ids; that out-of-vocabulary value is later consumed by the model's embedding and attention path and crashes the engine worker with a GPU device-side assertion. The same triggering request sequence is reachable through the public gRPC Generate and Abort endpoints, so a remote client that can send generation requests can crash the shared engine worker, aborting concurrent requests and causing a service-wide denial of service for other clients of the deployment until the worker is restarted. This issue is fixed in version 0.24.0.","aliases":["GHSA-8wr5-jm2h-8r4f"],"modified":"2026-07-09T04:02:06.068855414Z","published":"2026-07-06T19:49:20.481Z","database_specific":{"osv_generated_from":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/54xxx/CVE-2026-54234.json","cna_assigner":"GitHub_M","cwe_ids":["CWE-1284","CWE-20"]},"references":[{"type":"ADVISORY","url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/54xxx/CVE-2026-54234.json"},{"type":"ADVISORY","url":"https://github.com/vllm-project/vllm/security/advisories/GHSA-8wr5-jm2h-8r4f"},{"type":"ADVISORY","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-54234"},{"type":"FIX","url":"https://github.com/vllm-project/vllm/commit/8a5cf1ccd65e8ac7635c402c1ec0b08988bc26ca"},{"type":"FIX","url":"https://github.com/vllm-project/vllm/pull/44744"}],"affected":[{"ranges":[{"type":"GIT","repo":"https://github.com/vllm-project/vllm","events":[{"introduced":"0"},{"fixed":"217c64a976869883fcf0c52a8cf8bc3c954d285b"},{"fixed":"8a5cf1ccd65e8ac7635c402c1ec0b08988bc26ca"}],"database_specific":{"source":["CPE_RANGE","REFERENCES"],"extracted_events":[{"introduced":"0"},{"fixed":"0.24.0"}],"cpe":"cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*"}}],"versions":["v0.24.0rc2","v0.24.0rc1","v0.23.1rc0","v0.22.1rc0","v0.21.1rc0","v0.20.2rc0","v0.20.1rc0","v0.19.2rc0","v0.19.1rc0","v0.18.2rc0","v0.18.1rc0","v0.17.2rc0","v0.18.0rc0","v0.17.1rc0","v0.17.0rc0","v0.16.0rc1","v0.15.2rc0","v0.16.0rc0","v0.15.0rc1","v0.14.0rc1","v0.14.0rc0","v0.13.0rc1","v0.11.1","v0.11.1rc6","v0.11.1rc5","v0.11.1rc4","v0.11.1rc3","v0.11.1rc2","v0.11.1rc1","v0.11.1rc0","v0.11.0rc1","v0.10.2rc2","v0.10.2rc1","v0.10.1rc1","v0.10.0rc2","v0.10.0","v0.10.0rc1","v0.9.2rc1","v0.9.1rc2","v0.9.1","v0.9.1rc1","v0.9.0","v0.8.4","v0.8.3rc1","v0.8.2","v0.8.1","v0.8.0rc2","v0.8.0rc1","v0.7.3","v0.7.2","v0.7.1","v0.7.0","v0.6.6.post1","v0.6.6","v0.6.5","v0.6.4.post1","v0.6.4","v0.6.3.post1","v0.6.3","v0.6.2","v0.6.1.post2","v0.6.1.post1","v0.6.1","v0.6.0","v0.5.5","v0.5.4","v0.5.3.post1","v0.5.3","v0.5.2","v0.5.1","v0.5.0.post1","v0.5.0","v0.4.3","v0.4.2","v0.4.1","v0.4.0.post1","v0.4.0","v0.3.3","v0.3.2","v0.3.1","v0.3.0","v0.2.7","v0.2.6","v0.2.5","v0.2.4","v0.2.3","v0.2.2","v0.2.1","v0.2.0","v0.1.7","v0.1.6","v0.1.5","v0.1.4","v0.1.3","v0.1.2","v0.1.1","v0.1.0","submission"],"database_specific":{"source":"https://storage.googleapis.com/cve-osv-conversion/osv-output/CVE-2026-54234.json"}}],"schema_version":"1.7.5","severity":[{"type":"CVSS_V3","score":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H"}]}