Direct naar de inhoud
⚡ Kritieke cyberalerts voor jouw sector & systemen — direct in je inbox. Aanmelden →
cybernieuws.nl Cybersecurity en informatiebeveiling nieuws alerts live · 10 min 121 bronnen 1 kritiek 6 vandaag
CVE's Videos Dagbriefing

← Terug naar CVE-database

CVE-2026-71486

MEDIUM · 4.3 CVSS Gepubliceerd: CWE-400

Beschrijving NL

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.

Origineel (Engels) tonen

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.

Vendors

Vllm

Affected products

Vllm

References