CVE-2026-55574: CWE-1333: Inefficient Regular Expression Complexity in vllm-project vllm
vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, the structured_outputs.regex API parameter passes a user-supplied regular expression string directly to the grammar compiler backends with no compilation timeout; in the xgrammar backend the string reaches the regex compiler with no guard, and in the outlines backend the validation step blocks structural issues such as lookarounds and backreferences but performs no complexity analysis, so a pattern with nested quantifiers passes all checks and causes exponential state-space expansion, allowing a single request containing an adversarial regex to hang an inference worker indefinitely and deny service. This issue is fixed in version 0.24.0.
AI Analysis
Technical Summary
The vulnerability in vLLM (CVE-2026-55574) involves inefficient regular expression complexity handling in the structured_outputs.regex API parameter. Before version 0.24.0, user-supplied regex strings are forwarded to the grammar compiler backends without a compilation timeout. The xgrammar backend applies no guard, and the outlines backend only blocks certain structural issues but does not analyze regex complexity. Consequently, adversarial regex patterns with nested quantifiers can cause exponential expansion in the regex engine's state space, resulting in a denial of service by indefinitely hanging inference workers. This vulnerability is addressed in vLLM version 0.24.0.
Potential Impact
An attacker can supply a crafted regular expression that triggers exponential state-space expansion in the regex compiler, causing an inference worker to hang indefinitely. This results in a denial of service condition, impacting availability of the vLLM inference engine. The vulnerability requires no privileges or user interaction and can be exploited remotely via the structured_outputs.regex API parameter.
Mitigation Recommendations
Upgrade vLLM to version 0.24.0 or later, where this vulnerability is fixed by adding appropriate compilation timeouts and complexity checks on user-supplied regular expressions. No other mitigation is documented. Patch status is confirmed fixed in 0.24.0.
CVE-2026-55574: CWE-1333: Inefficient Regular Expression Complexity in vllm-project vllm
Description
vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, the structured_outputs.regex API parameter passes a user-supplied regular expression string directly to the grammar compiler backends with no compilation timeout; in the xgrammar backend the string reaches the regex compiler with no guard, and in the outlines backend the validation step blocks structural issues such as lookarounds and backreferences but performs no complexity analysis, so a pattern with nested quantifiers passes all checks and causes exponential state-space expansion, allowing a single request containing an adversarial regex to hang an inference worker indefinitely and deny service. This issue is fixed in version 0.24.0.
CVSS v4.0
Score 8.7high
Affected software
Run on your own infrastructure? Check whether these packages are installed with threat-finder — our free open-source scanner.
Weaknesses
AI-Powered Analysis
Machine-generated threat intelligence
Technical Analysis
The vulnerability in vLLM (CVE-2026-55574) involves inefficient regular expression complexity handling in the structured_outputs.regex API parameter. Before version 0.24.0, user-supplied regex strings are forwarded to the grammar compiler backends without a compilation timeout. The xgrammar backend applies no guard, and the outlines backend only blocks certain structural issues but does not analyze regex complexity. Consequently, adversarial regex patterns with nested quantifiers can cause exponential expansion in the regex engine's state space, resulting in a denial of service by indefinitely hanging inference workers. This vulnerability is addressed in vLLM version 0.24.0.
Potential Impact
An attacker can supply a crafted regular expression that triggers exponential state-space expansion in the regex compiler, causing an inference worker to hang indefinitely. This results in a denial of service condition, impacting availability of the vLLM inference engine. The vulnerability requires no privileges or user interaction and can be exploited remotely via the structured_outputs.regex API parameter.
Mitigation Recommendations
Upgrade vLLM to version 0.24.0 or later, where this vulnerability is fixed by adding appropriate compilation timeouts and complexity checks on user-supplied regular expressions. No other mitigation is documented. Patch status is confirmed fixed in 0.24.0.
Technical Details
- Data Version
- 5.2
- Assigner Short Name
- GitHub_M
- Date Reserved
- 2026-06-16T23:11:20.214Z
- Cvss Version
- 4.0
- State
- PUBLISHED
- Remediation Level
- null
Threat ID: 6a4c11fc27e9c797192ee541
Added to database: 07/06/2026, 20:37:16 UTC
Last enriched: 07/06/2026, 20:51:58 UTC
Last updated: 07/06/2026, 21:25:04 UTC
Views: 4
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.