CVE-2026-71486: CWE-400: Uncontrolled Resource Consumption in vllm-project vllm
vLLM versions prior to 0.26.0 contain a vulnerability in the /v1/completions/derender and /v1/chat/completions/derender endpoints where certain response structures are processed before enforcing limits on model length, tokens, sequences, or response size. This allows an authenticated API client to cause excessive CPU and memory consumption, resulting in oversized responses. The issue is addressed in version 0.26.0.
AI Analysis
Technical Summary
vLLM is an inference and serving engine for large language models. Before version 0.26.0, the /v1/completions/derender and /v1/chat/completions/derender endpoints accept GenerateResponse objects from authenticated clients. These objects include generate_responses, choices, token_ids, prompt_logprobs, logprobs.content, top_logprobs, and routed_experts, which are processed by OnlineDerenderer and tokenizer.decode prior to enforcement of limits such as max_model_len, max_tokens, max_num_seqs, or response size. This sequence allows an attacker to consume excessive CPU and memory resources and generate oversized responses, constituting uncontrolled resource consumption (CWE-400) and improper restriction of operations within resource limits (CWE-770). The vulnerability is fixed in vLLM version 0.26.0.
Potential Impact
An authenticated API client can exploit this vulnerability to cause excessive CPU and memory usage on the server, potentially leading to denial of service or degraded performance. There is no direct impact on confidentiality or integrity according to the CVSS vector. The severity is rated medium with a CVSS score of 4.3.
Mitigation Recommendations
Upgrade vLLM to version 0.26.0 or later, where this vulnerability is fixed. No other official remediation or temporary fixes are documented. Patch status is confirmed fixed in 0.26.0.
CVE-2026-71486: CWE-400: Uncontrolled Resource Consumption in vllm-project vllm
Description
vLLM versions prior to 0.26.0 contain a vulnerability in the /v1/completions/derender and /v1/chat/completions/derender endpoints where certain response structures are processed before enforcing limits on model length, tokens, sequences, or response size. This allows an authenticated API client to cause excessive CPU and memory consumption, resulting in oversized responses. The issue is addressed in version 0.26.0.
CVSS v3.1
Score 4.3medium
Affected software
Run on your own infrastructure? Check whether these packages are installed with threat-finder — our free open-source scanner.
AI-Powered Analysis
Machine-generated threat intelligence
Technical Analysis
vLLM is an inference and serving engine for large language models. Before version 0.26.0, the /v1/completions/derender and /v1/chat/completions/derender endpoints accept GenerateResponse objects from authenticated clients. These objects include generate_responses, choices, token_ids, prompt_logprobs, logprobs.content, top_logprobs, and routed_experts, which are processed by OnlineDerenderer and tokenizer.decode prior to enforcement of limits such as max_model_len, max_tokens, max_num_seqs, or response size. This sequence allows an attacker to consume excessive CPU and memory resources and generate oversized responses, constituting uncontrolled resource consumption (CWE-400) and improper restriction of operations within resource limits (CWE-770). The vulnerability is fixed in vLLM version 0.26.0.
Potential Impact
An authenticated API client can exploit this vulnerability to cause excessive CPU and memory usage on the server, potentially leading to denial of service or degraded performance. There is no direct impact on confidentiality or integrity according to the CVSS vector. The severity is rated medium with a CVSS score of 4.3.
Mitigation Recommendations
Upgrade vLLM to version 0.26.0 or later, where this vulnerability is fixed. No other official remediation or temporary fixes are documented. Patch status is confirmed fixed in 0.26.0.
Technical Details
- Data Version
- 5.2
- Assigner Short Name
- GitHub_M
- Date Reserved
- 2026-08-06T19:56:23.725Z
- Cvss Version
- 3.1
- State
- PUBLISHED
- Remediation Level
- null
Threat ID: 6a836e8fbf8831d5398572cf
Added to database: 08/17/2026, 20:26:55 UTC
Last enriched: 08/17/2026, 20:45:41 UTC
Last updated: 08/18/2026, 00:05:51 UTC
Views: 7
Community Reviews
0 reviewsCrowdsource mitigation strategies, share intel context, and vote on the most helpful responses. Sign in to add your voice and help keep defenders ahead.
Want to contribute mitigation steps or threat intel context? Sign in or create an account to join the community discussion.
Actions
Updates to AI analysis require Pro Console access. Upgrade inside Console → Billing.
Need more coverage?
Upgrade to Pro Console for AI refresh and higher limits.
For incident response and remediation, OffSeq services can help resolve threats faster.
Latest Threats
Check if your credentials are on the dark web
Instant breach scanning across billions of leaked records. Free tier available.