{"id":"CVE-2026-71486","title":"vLLM is an inference and serving engine for large language models","summary":"vLLM is an inference and serving engine for large language models. Prior to 0.26.0, the /v1/completions/derender and /v1/chat/completions/derender endpoints accept caller-supplied GenerateResponse objects whose generate_responses, choice…","severity":"medium","cvss":4.3,"cvssVector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:L","cwe":["CWE-400","CWE-770"],"vendor":"vllm","product":"vllm","affected":["vllm < 0.26.0"],"patched":["vllm 0.26.0"],"published":"2026-08-17","updated":"2026-09-09","sourceUpdated":"2026-09-09T21:11:25.743","source":"NVD","sourceUrl":"https://nvd.nist.gov/vuln/detail/CVE-2026-71486","references":[{"url":"https://github.com/vllm-project/vllm/commit/8e61b646e2d157f9b93451fa048f9c8530c8a67b","label":"security-advisories@github.com"},{"url":"https://github.com/vllm-project/vllm/pull/47260","label":"security-advisories@github.com"},{"url":"https://github.com/vllm-project/vllm/releases/tag/v0.26.0","label":"security-advisories@github.com"},{"url":"https://github.com/vllm-project/vllm/security/advisories/GHSA-8737-qx52-hjff","label":"security-advisories@github.com"},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-71486"},{"url":"https://github.com/advisories/GHSA-8737-qx52-hjff"},{"url":"https://github.com/vllm-project/vllm"},{"url":"https://pypi.org/project/vllm"}],"tags":["nvd","ghsa","pip","osv"],"epss":0.00341,"epssPercentile":0.27765,"aliases":["GHSA-8737-qx52-hjff","PYSEC-2026-3936"],"ecosystem":"pip","ingestedAt":"2026-09-04T22:28:54.993Z","slug":"CVE-2026-71486","body":"## Overview\n\nvLLM is an inference and serving engine for large language models. Prior to 0.26.0, the /v1/completions/derender and /v1/chat/completions/derender endpoints accept caller-supplied GenerateResponse objects whose generate_responses, choices, token_ids, prompt_logprobs, logprobs.content, top_logprobs, and routed_experts structures are processed by OnlineDerenderer and tokenizer.decode before max_model_len, max_tokens, max_num_seqs, or response-size limits are enforced, allowing an authenticated API client to consume excessive CPU and memory and produce oversized responses. This issue is fixed in version 0.26.0.\n\n## Remediation\n\nRefer to the linked advisories for vendor-supplied fixes and affected version ranges.\n\n## Package advisory (CVE-2026-71486)\n\nAffected packages:\n\n- `vllm < 0.26.0`\n\nPatched in:\n\n- `vllm 0.26.0`\n\nSource: https://github.com/advisories/GHSA-8737-qx52-hjff","depth":"sunlit","depthScore":24,"depthScoreParts":{"impact":23.7,"likelihood":0.1,"exploitation":0,"ransomware":0},"changes":[]}