CVE-2026-33626 | LMDeploy Vulnerable to Server-Side Request Forgery (SSRF) via Vision-Language Image Loading

Exp

LMDeploy is a toolkit for compressing, deploying, and serving large language models. Versions prior to 0.12.3 have a Server-Side Request Forgery (SSRF) vulnerability in LMDeploy's vision-language module. The `load_image()` function in `lmdeploy/vl/utils.py` fetches arbitrary URLs without validating internal/private IP addresses, allowing attackers to access cloud metadata services, internal networks, and sensitive resources. Version 0.12.3 patches the issue.

Published: 2026-04-20 Last update: 2026-06-17 Assigner: [email protected] Source: [email protected]

Conclusion & alert: CVE-2026-33626 is rated High Exploit Risk (84.5/100): CVSS High severity, with high exploitation likelihood (EPSS 45.25%, 99th percentile). Core evidence: 1 public exploit reference(s) are indexed (Exploit-DB). EPSS rose +37.26% over the last day, indicating growing attacker interest. Mandatory action: Public exploits are available—assess exposure, apply mitigations, and prioritize patching.

Risk is dynamic; we continuously reassess and refresh what is shown on this page as upstream context changes.

Public exploit references (Exploit-DB) for CVE-2026-33626

EDB-ID Source Kind Published Link
nvd_ref exploit_tag Exploit-DB ↗

Exploit prediction scoring system (EPSS) score for CVE-2026-33626

EPSS lead: Daily EPSS estimates relative likelihood of exploitation; percentile ranks this CVE among scored vulnerabilities (higher = more severe relative rank).

# Date Old EPSS score New EPSS score Delta (New - Old)
1 2026-06-15 7.99% 45.25% +37.26%
2 2026-06-14 8.70% 7.99% -0.71%
3 2026-05-22 8.70%

Full EPSS history (5 records total)

Common vulnerability scoring system (CVSS) metrics for CVE-2026-33626

CVSS metrics for this CVE.

Base score Version Severity Vector Exploitability Impact Score source
7.5 3.1 HIGH
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N Click to expand
Attack vector (AV:N)
Could be attacked over the internet or any normal routed network—not just someone sitting at the machine.
Attack complexity (AC:L)
Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:N)
No account or special rights needed—anonymous or random user is enough.
User interaction (UI:N)
Nobody has to click “OK” or open a trap file; it can work without a victim helping.
Scope (S:U)
Damage stays in the same “trust bubble” as the broken component—no big spill into unrelated systems.
Confidentiality (C:H)
Serious risk that confidential data gets exposed in a big way.
Integrity (I:N)
Data isn’t meaningfully altered or forged.
Availability (A:N)
Service keeps running; no real outage angle.
3.9 3.6 [email protected]

Weakness enumeration for CVE-2026-33626

GitHub Security Advisory for CVE-2026-33626

GHSA-6w67-hwm5-92mq · Severity: high · Ecosystem: pip — LMDeploy has Server-Side Request Forgery (SSRF) via Vision-Language Image Loading

Affected software / configurations for CVE-2026-33626

Vendor Product Version Raw CPE
internlm lmdeploy < 0.12.3 cpe:2.3:a:internlm:lmdeploy:*:*:*:*:*:*:*:*

References for CVE-2026-33626

cvelogic Threat Intelligence