CVE-2026-71486 | BDVDB-2026-360629 Detail

Description#

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, 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.

Impact#

Weaknesses#

References#

Exploit#

History#

1. Version 12026-08-17Record published in the CVE List.
2. Version 22026-08-18Most recent update recorded in the CVE List.