Description
LMDeploy is a toolkit for compressing, deploying, and serving large language models. From 0.12.3 until 0.15.0, _is_safe_url validates remote media URLs with urlparse while requests.Session().get interprets backslashes differently when sending the request. An attacker can provide a URL whose validation hostname appears globally routable but whose request destination resolves to an internal host, bypassing non-global address checks. Applications that accept attacker-controlled remote media URLs can therefore be induced to access internal HTTP services and expose data available from those services. This issue is fixed in version 0.15.0.
Published: 2026-11-09
Score: 7.5 High
EPSS: n/a
KEV: No
Impact: Remote Server Side Request Forgery
Action: Immediate Patch
AI Analysis

Impact

The vulnerability occurs when the toolkit validates remote media URLs using a parser that treats backslashes differently than the HTTP client, allowing a URL that appears to point to a public hostname to resolve to an internal host after the request is sent. This bypasses the non‑global address checks and permits an attacker to cause the application to access internal HTTP services, potentially exposing sensitive data.

Affected Systems

InternLM lmdeploy products from version 0.12.3 up to, but not including, version 0.15.0 are affected. The issue was fixed in release 0.15.0.

Risk and Exploitability

With a CVSS score of 7.5, the vulnerability is considered high severity, yet EPSS information is not available and it is not listed in the CISA KEV catalog. The likely attack path involves an attacker‑controlled media URL that the application accepts; the URL is then interpreted by the HTTP client and resolved to an internal host, enabling unauthorized data exfiltration from internal services.

Generated by OpenCVE AI on November 9, 2026 at 22:35 UTC.

Remediation

No solution or workaround provided in the CVE record.

OpenCVE Recommended Actions

  • Upgrade InternLM lmdeploy to version 0.15.0 or later.
  • If upgrading is currently infeasible, restrict the media URL configuration to a strict whitelist of trusted domains or prohibit any HTTP/HTTPS URLs from untrusted input.
  • Implement network segmentation or firewall controls that block outbound connections from the application to internal IP ranges, reducing the impact of a successful SSRF.

Generated by OpenCVE AI on November 9, 2026 at 22:35 UTC.

Tracking

Sign in to view the affected projects.

Advisories
Source ID Title
Github GHSA Github GHSA GHSA-39wr-7q6h-cf68 LMDeploy has an SSRF bypass
History

Mon, 09 Nov 2026 22:30:00 +0000

Type Values Removed Values Added
First Time appeared Internlm
Internlm lmdeploy
Vendors & Products Internlm
Internlm lmdeploy

Mon, 09 Nov 2026 21:30:00 +0000

Type Values Removed Values Added
Description LMDeploy is a toolkit for compressing, deploying, and serving large language models. From 0.12.3 until 0.15.0, _is_safe_url validates remote media URLs with urlparse while requests.Session().get interprets backslashes differently when sending the request. An attacker can provide a URL whose validation hostname appears globally routable but whose request destination resolves to an internal host, bypassing non-global address checks. Applications that accept attacker-controlled remote media URLs can therefore be induced to access internal HTTP services and expose data available from those services. This issue is fixed in version 0.15.0.
Title LMDeploy URL parser disagreement permits SSRF filter bypass
Weaknesses CWE-918
References
Metrics cvssV3_1

{'score': 7.5, 'vector': 'CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N'}


Subscriptions

Internlm Lmdeploy
cve-icon MITRE

Status: PUBLISHED

Assigner: GitHub_M

Published:

Updated: 2026-11-09T21:19:51.856Z

Reserved: 2026-11-06T15:27:56.227Z

Link: CVE-2026-123818

cve-icon Vulnrichment

No data.

cve-icon NVD

No data.

cve-icon Redhat

No data.

cve-icon OpenCVE Enrichment

Updated: 2026-11-09T22:15:15Z

Weaknesses
  • CWE-918

    Server-Side Request Forgery (SSRF)