Vulnerability PYSEC-2026-3998
Medium Risk
MEDIUM RISK
CVSS Score: 5.3
Score Range: 4.0–6.9
Medium severity vulnerabilities (CVSS 4.0–6.9). Important issues that meaningfully reduce security confidence.
10 days ago
September 18, 2026 at 08:17 PM UTC
No summary available
0.0.1 - 0.28.0
0.0.1 - 0.28.0
Details
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.
Related Vulnerabilities
Other vulnerabilities affecting the same packages
Medium Risk
9 days ago
No summary available
0.0.1 - 0.29.0 PYSEC-2026-4000
0.0.1 - 0.29.0 PYSEC-2026-4000
Medium Risk
10 days ago
No summary available
0.0.1 - 0.29.0 PYSEC-2026-3999
0.0.1 - 0.29.0 PYSEC-2026-3999
Unknown
10 days ago
No summary available
0.0.1 - 0.27.1 PYSEC-2026-3997
0.0.1 - 0.27.1 PYSEC-2026-3997
Unknown
11 days ago
No summary available
0.0.1 - 0.29.0 PYSEC-2026-3996
0.0.1 - 0.29.0 PYSEC-2026-3996
Medium Risk
11 days ago
vLLM: Request-selected PyNvVideoCodec GPU decode bypasses static VRAM reservation
0.0.1 - 0.27.1 GHSA-8pw2-6jv3-mj5j
0.0.1 - 0.27.1 GHSA-8pw2-6jv3-mj5j
Impacted packages
Timeline
Published
10 days ago
September 18, 2026 at 08:17 PM UTC
Fixed (0.29.0)
20 days ago
September 09, 2026 at 08:58 AM UTC
Last Modified
2 hours ago
September 29, 2026 at 09:00 AM UTC