CVE-2026-93989
vLLM through 0.29.0 Cross-Request Logits Corruption via bad_words
CVSS Score
3.1
EPSS Score
0.0%
EPSS Percentile
0th
vLLM through 0.29.0 fails to properly validate bad_words token indices against the model's generation output width in SamplingParams.update_from_tokenizer(). Attackers can supply out-of-bounds token indices that corrupt logits memory of concurrent requests, causing different in-flight HTTP requests to return incorrect tokens.
| CWE | CWE-129 |
| Vendor | vllm-project |
| Product | vllm |
| Published | Sep 19, 2026 |
Stay Ahead of the Next One
Get instant alerts for vllm-project vllm
Be the first to know when new low vulnerabilities affecting vllm-project vllm are published β delivered to Slack, Telegram or Discord.
Get Free Alerts β
Free Β· No credit card Β· 60 sec setup
CVSS v3 Breakdown
CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:N Attack Vector
Network
Attack Complexity
High
Privileges Required
Low
User Interaction
None
Scope
Unchanged
Confidentiality
None
Integrity
Low
Availability
None
Affected Versions
vllm-project / vllm
0 β€ 0.29.0
References
github.com: https://github.com/vllm-project/vllm/pull/48824 github.com: https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/sample/bad_words.py github.com: https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/sampling_params.py#L694-L753 github.com: https://github.com/vllm-project/vllm vulncheck.com: https://www.vulncheck.com/advisories/vllm-through-0.29.0-cross-request-logits-corruption-via-bad-words
Credits
π δΊζζ·