vllm before 0.29.0 Denial of Service via Decoder Prompt Length Bypass
vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate. When the request contains a 'features' (multimodal) payload, vllm/entrypoints/serve/disagg/serving.py builds a multimodal EngineInput directly from the caller-supplied token_ids, and Genera…
Severity high: primary CNA CVSS 7.1 v4.0 (worst CNA-provided scenario) Not listed in this KEV snapshot. Forecast present in this snapshot.
Published 2026-09-26T13:23:21.141Z. Updated 2026-09-28T18:25:40.222Z.