CVE-2026-73556CWE-1333CWE-400denial-of-service

CVE-2026-73556

Medium · published August 13, 2026

CVSS v3.1
5.3
EPSS
1%
Percentile
43.0
In the wild
Unconfirmed
What it is

vLLM is an inference and serving engine for large language models. Prior to 0.26.0, the structured_outputs.regex parameter in vllm/v1/structured_output/backend_lm_format_enforcer.py is passed to lmformatenforcer.RegexParser without compile_regex_with_timeout or validation in validate_structured_output_request_lm_format_enforcer, allowing an unauthenticated /v1/completions request against the lm-format-enforcer backend to consume a CPU core and stall the structured-output engine path with a catastrophic regular expression. This issue is fixed in version 0.26.0.

The record
Technical detail
CVSS v3.1
5.3 · MEDIUM
Vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L
CVSS v4.0
Not supplied
EPSS
0.00515 · 43.0th percentile
Weaknesses
CWE-1333 · Inefficient Regular Expression Complexity; CWE-400 · Uncontrolled Resource Consumption
Published
2026-08-13T19:20Z
Affected products (1)
ProductVersionsFixed in
vllm/vllm< 0.26.00.26.0
References (5)
EPSS history
Timeline
  • 13 AUG 14:56Z
    vLLM: ReDoS via structured_outputs.regex in the lm-format-enforcer backend (no compile timeout) — missed sibling of CVE-2026-55574
    cvelistv5