CVE-2026-93840

LOWCVSS 3.7/10

Last modified

CVE-2026-93840 is a low-severity vulnerability rated 3.7/10 on the CVSS scale. vLLM before 0.29.0 validates allowed_token_ids against tokenizer length instead of model output logits width in SamplingParams._validate_allowed_token_ids(). Attackers can supply token IDs above the output vocabulary that pass validation, causing LogitBiasState to corrupt GPU logits state and allow concurrent requests to sample tokens outside their allowlists..

Description

vLLM before 0.29.0 validates allowed_token_ids against tokenizer length instead of model output logits width in SamplingParams._validate_allowed_token_ids(). Attackers can supply token IDs above the output vocabulary that pass validation, causing LogitBiasState to corrupt GPU logits state and allow concurrent requests to sample tokens outside their allowlists.

Metrics

Weakness Enumeration

Affected Software

Source: CNA advisory (CVE.org). NVD analysis pending.

VendorProductVersions
vllm-projectvllm< 0.29.0

References

Timeline

Published
Last Modified
Status
Received

Frequently Asked Questions

What is CVE-2026-93840?
vLLM before 0.29.0 validates allowed_token_ids against tokenizer length instead of model output logits width in SamplingParams._validate_allowed_token_ids(). Attackers can supply token IDs above the output vocabulary that pass validation, causing LogitBiasState to corrupt GPU logits state and allow concurrent requests to sample tokens outside their allowlists.
How severe is CVE-2026-93840?
CVE-2026-93840 has a CVSS score of 3.7/10 (LOW severity).
How do I fix CVE-2026-93840?
Check the vendor references and advisories linked above for patched versions and mitigation guidance. You can also run a Strix scan to test if your systems are affected.

How Strix Helps

Related CVEs from 2026

Are you affected by CVE-2026-93840?

Run a free Strix scan to check your systems for this vulnerability.

Scan your code now

Source: NVD / NIST