CVE-2026-54234
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…
Does this matter?
High impact if exploited, but EPSS currently rates exploitation as unlikely (0.62%). Schedule it in the normal patch cycle and watch for a rise in EPSS or a public exploit.
Description
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.
- CVSS 3.1
- 7.5 HIGHCVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
- EPSS
- 0.62% probability · 48th percentile
- CISA KEV
- Not listed
- Weakness
- CWE-20, CWE-1284
- Affected
- vllm/vllm
- Source
- security-advisories@github.com
References
- https://github.com/vllm-project/vllm/commit/8a5cf1ccd65e8ac7635c402c1ec0b08988bc26caPatch
- https://github.com/vllm-project/vllm/pull/44744Issue Tracking, Patch
- https://github.com/vllm-project/vllm/security/advisories/GHSA-8wr5-jm2h-8r4fVendor Advisory, Exploit
- https://github.com/vllm-project/vllm/security/advisories/GHSA-8wr5-jm2h-8r4fVendor Advisory, Exploit
Source: NVD record, EPSS from FIRST.org, KEV from CISA. Refreshed daily.