CVE-2025-46570

Summary

vLLM is an inference and serving engine for large language models (LLMs). Prior to version 0.9.0, when a new prompt is processed, if the PageAttention mechanism finds a matching prefix chunk, the prefill process speeds up, which is reflected in the TTFT (Time to First Token). These timing difference

Severity

medium

EPSS

0.2% (p16)

Also known as

GHSA-4qjh-9fv9-r85r, PYSEC-2025-53

Published

2025-05-29

References

• https://github.com/vllm-project/vllm/pull/17045
• https://github.com/vllm-project/vllm/security/advisories/GHSA-4qjh-9fv9-r85r
• https://github.com/vllm-project/vllm/commit/77073c77bc2006eb80ea6d5128f076f5e6c6f54f

Related advisories

• CVE-2026-48746 — critical · PyPI/vllm
• CVE-2026-22778 — critical · PyPI/vllm
• CVE-2025-47277 — critical · PyPI/vllm
• CVE-2025-32444 — critical · PyPI/vllm
• GHSA-ggpf-24jw-3fcw — critical · PyPI/vllm
• CVE-2024-9052 — critical · PyPI/vllm
• CVE-2024-9053 — critical · PyPI/vllm
• CVE-2024-11041 — critical · PyPI/vllm

Is your project exposed to this? Stateward checks every dependency on every pull request and flags it only if your code actually reaches it.

Summarize with AI

ChatGPTClaudePerplexity