CVE-2026-105754

vLLM is an inference and serving engine for large language models. Prior to 0.30.0, the /inference/v1/generate endpoint in the disaggregated scale-out path accepts caller-supplied tensors in the features.kwargs_data field, cache identifiers in the features.mm_hashes field, ranges in the features.mm_placeholders field, and wire-selected multimodal field processors without rebinding them to the active model renderer contract. Forged grid geometry, field types, or non-positive placeholder lengths can terminate the shared EngineCore; when an attacker knows or can induce a victim's content hash, forged cache hashes can poison or retrieve cross-request encoder-cache state; and dropped sparse placeholder masks can alter replayed transport semantics. This issue is fixed in version 0.30.0.

• https://github.com/vllm-project/vllm/commit/1970f3ed4be7fa8620e4ddc4a12c36a8384cfc27Patch
• https://github.com/vllm-project/vllm/pull/51898Issue Tracking
• https://github.com/vllm-project/vllm/releases/tag/v0.30.0Release Notes
• https://github.com/vllm-project/vllm/security/advisories/GHSA-ph72-cqr5-qpp7Mitigation