vLLM is an inference and serving engine for large language models. Prior to 0.32.0, vllm/tokenizers/hf.py computes max_token_id with an off-by-one upper bound, allowing a pre-tokenized prompt ID equal to the vocabulary size to pass validation. For a tensor-parallel size of one and an exact multiple-of-64 vocabulary, VocabParallelEmbedding.forward does not mask that value before F.embedding uses it, causing a CUDA device-side assertion and terminating the engine. This issue is fixed in version 0.32.0.
{
"cna_assigner": "GitHub_M",
"cwe_ids": [
"CWE-129",
"CWE-193"
],
"osv_generated_from": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/115xxx/CVE-2026-115916.json"
}{
"extracted_events": [
{
"cpe": "cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*",
"range": [
{
"introduced": "0"
},
{
"fixed": "0.32.0"
}
],
"source": "CPE_RANGE"
},
{
"range": [
{
"fixed": "37cecc8141e2dcba9dd534897a164d4baaf27229"
}
],
"source": "REFERENCES_COMMIT"
},
{
"original_tag": "v0.32.0",
"range": [
{
"fixed": "5df63a7203cdb9328d824a27b682e77c7b9928be"
}
],
"source": "REFERENCES_TAG"
}
]
}