CVE-2026-115916
Medium
Description
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.
SeverityMedium
CVSS6.5
Exploit probability—
ProductVllm
PublishedOct 26, 2026
First seenOct 26, 2026
Weakness typeCWE-129CWE-129CWE-193CWE-193
How to fix
Update
- vllm vllm0.32.0 or later
- vllm0.32.0 or later
Affected products
| Product | Affected versions | Fixed in |
|---|---|---|
| vllm vllm | v < 0.32.0 | 0.32.0 |
| vllmPyPI | v < 0.32.0 | 0.32.0 |
Check your version
Choose the product and enter the version you run to see whether it falls in the affected range.
In the Intelligence 360 plan
Attack Surface Management
See the assets tied to your domains, the technologies they run and known vulnerabilities in one list.