Lewati ke konten
CodeAuditAgent

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.

Kelemahan

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.

Referensi

Temukan bug sebelum penyerang menemukannya.

Masuk dengan GitHub dan jalankan audit pertama Anda dalam waktu kurang dari satu menit. Paket gratis tidak memerlukan kartu kredit.