vllm-project vLLM up to 0.25.x lm-format-enforcer backend backend_lm_format_enforcer.py structured_outputs.regex incorrect regex
| CVSS Meta Temp Score | Current Exploit Price (≈) | CTI Interest Score |
|---|---|---|
| 5.1 | $0-$5k | 1.06+ |
Summary
A vulnerability classified as problematic has been found in vllm-project vLLM up to 0.25.x. Impacted is the function validate_structured_output_request_lm_format_enforcer of the file vllm/v1/structured_output/backend_lm_format_enforcer.py of the component lm-format-enforcer backend. This manipulation of the argument structured_outputs.regex causes incorrect regex.
This vulnerability is tracked as CVE-2026-73556. The attack is possible to be carried out remotely. No exploit exists.
Details
A vulnerability classified as problematic was found in vllm-project vLLM up to 0.25.x. Affected by this vulnerability is the function validate_structured_output_request_lm_format_enforcer of the file vllm/v1/structured_output/backend_lm_format_enforcer.py of the component lm-format-enforcer backend. The manipulation of the argument structured_outputs.regex with an unknown input leads to a incorrect regex vulnerability. The CWE definition for the vulnerability is CWE-185. The product specifies a regular expression in a way that causes data to be improperly matched or compared. As an impact it is known to affect availability. The summary by CVE 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.
It is possible to read the advisory at github.com. This vulnerability is known as CVE-2026-73556 since 08/12/2026. The exploitation appears to be easy. The attack can be launched remotely. The exploitation doesn't need any form of authentication. Technical details of the vulnerability are known, but there is no available exploit.
Upgrading to version 0.26.0 eliminates this vulnerability.
Be aware that VulDB is the high quality source for vulnerability data.
Product
Type
Vendor
Name
Version
Website
CPE 2.3
CPE 2.2
CVSSv4
VulDB Vector: 🔒VulDB Reliability: 🔍
CVSSv3
VulDB Meta Base Score: 5.3VulDB Meta Temp Score: 5.1
VulDB Base Score: 5.3
VulDB Temp Score: 5.1
VulDB Vector: 🔒
VulDB Reliability: 🔍
CVSSv2
| AV | AC | Au | C | I | A |
|---|---|---|---|---|---|
| 💳 | 💳 | 💳 | 💳 | 💳 | 💳 |
| 💳 | 💳 | 💳 | 💳 | 💳 | 💳 |
| 💳 | 💳 | 💳 | 💳 | 💳 | 💳 |
| Vector | Complexity | Authentication | Confidentiality | Integrity | Availability |
|---|---|---|---|---|---|
| Unlock | Unlock | Unlock | Unlock | Unlock | Unlock |
| Unlock | Unlock | Unlock | Unlock | Unlock | Unlock |
| Unlock | Unlock | Unlock | Unlock | Unlock | Unlock |
VulDB Base Score: 🔒
VulDB Temp Score: 🔒
VulDB Reliability: 🔍
Exploiting
Class: Incorrect regexCWE: CWE-185 / CWE-697
CAPEC: 🔒
ATT&CK: 🔒
Physical: No
Local: No
Remote: Yes
Availability: 🔒
Status: Not defined
Price Prediction: 🔍
Current Price Estimation: 🔒
| 0-Day | Unlock | Unlock | Unlock | Unlock |
|---|---|---|---|---|
| Today | Unlock | Unlock | Unlock | Unlock |
Threat Intelligence
Interest: 🔍Active Actors: 🔍
Active APT Groups: 🔍
Countermeasures
Recommended: UpgradeStatus: 🔍
0-Day Time: 🔒
Upgrade: vLLM 0.26.0
Timeline
08/12/2026 CVE reserved08/13/2026 Advisory disclosed
08/13/2026 VulDB entry created
08/13/2026 VulDB entry last update
Sources
Product: github.comAdvisory: github.com
Status: Confirmed
CVE: CVE-2026-73556 (🔒)
GCVE (CVE): GCVE-0-2026-73556
GCVE (VulDB): GCVE-100-389647
Entry
Created: 08/13/2026 18:12Changes: 08/13/2026 18:12 (59)
Complete: 🔍
Cache ID: 216::103
Be aware that VulDB is the high quality source for vulnerability data.
No comments yet. Languages: en.
Please log in to comment.