CVE-2025-24357 | vLLM allows a malicious model RCE by torch.load in hf_model_weights_iterator

vLLM is a library for LLM inference and serving. vllm/model_executor/weight_utils.py implements hf_model_weights_iterator to load the model checkpoint, which is downloaded from huggingface. It uses the torch.load function and the weights_only parameter defaults to False. When torch.load loads malicious pickle data, it will execute arbitrary code during unpickling. This vulnerability is fixed in v0.7.0.

Published: 2025-01-27 Last update: 2025-06-27 Assigner: [email protected] Source: [email protected]
NVD Status：Analyzed，CVE State: published
View at NVD, CVE.org, EUVD

CVE-2025-24357 is rated Moderate Risk (46.1/100): CVSS High severity, with low exploitation likelihood (EPSS 0.65%). Review affected assets and schedule remediation.

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

Exploit prediction scoring system (EPSS) score for CVE-2025-24357

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 0.66% 0.65% -0.02%
2 2026-06-09 0.90% 0.66% -0.24%
3 2026-06-06 — 0.90% —

Full EPSS history (13 records total)

Common vulnerability scoring system (CVSS) metrics for CVE-2025-24357

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:H/PR:N/UI:R/S:U/C:H/I:H/A:H 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:H)
Even with access, the exploit needs extra luck, timing, or a fussy environment to actually work.
Privileges required (PR:N)
No account or special rights needed—anonymous or random user is enough.
User interaction (UI:R)
A real person has to do something—click, install, enable—otherwise it doesn’t land.
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:H)
They could widely tamper with or forge data—trust in the data is badly hurt.
Availability (A:H)
Could take the service down hard or make it unusable for people who depend on it.
 1.6 5.9 [email protected]
8.8 3.1 HIGH
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H 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:R)
A real person has to do something—click, install, enable—otherwise it doesn’t land.
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:H)
They could widely tamper with or forge data—trust in the data is badly hurt.
Availability (A:H)
Could take the service down hard or make it unusable for people who depend on it.
 2.8 5.9 [email protected]

Weakness enumeration for CVE-2025-24357

GitHub Security Advisory for CVE-2025-24357

GHSA-rh4j-5rhw-hr54 · Severity: high · Ecosystem: pip — vllm: Malicious model to RCE by torch.load in hf_model_weights_iterator

OS Trackers for CVE-2025-24357

vendor priority summary link
redhat high — https://access.redhat.com/security/cve/CVE-2025-24357

Affected software / configurations for CVE-2025-24357

Vendor Product Version Raw CPE
vllm vllm < 0.7.0 cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*

References for CVE-2025-24357

cvelogic Threat Intelligence