vLLM allows Remote Code Execution by Pickle Deserialization via AsyncEngineRPCServer() RPC server entrypoints

Description

References

• nvd.nist.gov/vuln/detail/CVE-2024-9053 · ADVISORY
• pypa/advisory-database · WEB
• vllm-project/vllm · PACKAGE
• huntr.com/bounties/75a544f3-34a3-4da0-b5a3-1495cb031e09 · WEB
• GitHub Advisory GHSA-cj47-qj6g-x7r4 · ADVISORY

vllm 的其他漏洞

• CVE-2026-115918中危 6.5vLLM is an inference and serving engine for large language models. From 0.16.0 until 0.32.0, the /v1/responses endpoint allows an authentica…
• CVE-2026-115917中危 4.7vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the OpenAI-compatible server configures Starlette CORSMi…
• CVE-2026-115916中危 6.5vLLM is an inference and serving engine for large language models. Prior to 0.32.0, vllm/tokenizers/hf.py computes max_token_id with an off-…
• CVE-2026-115915严重 9.8vLLM is an inference and serving engine for large language models. Prior to 0.32.0, ImageMediaIO passes attacker-controlled image bytes to P…
• CVE-2026-115914中危 6.5vLLM is an inference and serving engine for large language models. From 0.11.1 until 0.32.0, the DeepSeek-OCR, DeepSeek-OCR-2, and Unlimited…