2026-09-26 13:23CVE-2026-100650VulnCheck
PUBLISHED5.2CWE-400

vLLM before 0.29.0 Resource Exhaustion via Unbounded Media Materialization

vLLM through 0.29.0 fetches and fully materializes remote or inline media before enforcing its documented media controls (the VLLM_MAX_AUDIO_CLIP_FILESIZE_MB compressed-audio size cap, default 25 MB, and the per-modality --limit-mm-per-prompt item limits). Across four ingress paths — the shared media-acquisition layer (HTTPConnection.get_bytes()/async_get_bytes()), the chat completions audio_url/base64 path, the batch speech runner, and the Rust frontend POST /tokenize route — the server reads the entire HTTP response body, base64-decodes the inline payload, or spawns one fetch/decode task per media part, and only then applies the limit (or, on some paths, never applies it). A remote attacker can therefore cause the API server or batch-runner process to allocate memory and consume outbound bandwidth proportional to an attacker-chosen body size or media item count before the request is rejected, resulting in pre-inference memory and bandwidth exhaustion (denial of service). The chat and batch surfaces require an API key when one is configured; the Rust frontend /tokenize route is unauthenticated by design. There is no code execution or data disclosure impact.

Problem type

Affected products

vllm-project

vllm

< 0.29.0 - AFFECTED

0.29.0 - UNAFFECTED

References

GitHub Security Advisories

GHSA-j8gm-f3cq-6c27

vLLM through 0.29.0 fetches and fully materializes remote or inline media before enforcing its...

https://github.com/advisories/GHSA-j8gm-f3cq-6c27

vLLM through 0.29.0 fetches and fully materializes remote or inline media before enforcing its documented media controls (the VLLM_MAX_AUDIO_CLIP_FILESIZE_MB compressed-audio size cap, default 25 MB, and the per-modality --limit-mm-per-prompt item limits). Across four ingress paths — the shared media-acquisition layer (HTTPConnection.get_bytes()/async_get_bytes()), the chat completions audio_url/base64 path, the batch speech runner, and the Rust frontend POST /tokenize route — the server reads the entire HTTP response body, base64-decodes the inline payload, or spawns one fetch/decode task per media part, and only then applies the limit (or, on some paths, never applies it). A remote attacker can therefore cause the API server or batch-runner process to allocate memory and consume outbound bandwidth proportional to an attacker-chosen body size or media item count before the request is rejected, resulting in pre-inference memory and bandwidth exhaustion (denial of service). The chat and batch surfaces require an API key when one is configured; the Rust frontend /tokenize route is unauthenticated by design. There is no code execution or data disclosure impact.

JSON source

https://cveawg.mitre.org/api/cve/CVE-2026-100650
Click to expand
{
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2",
  "cveMetadata": {
    "cveId": "CVE-2026-100650",
    "assignerOrgId": "83251b91-4cc7-4094-a5c7-464a1b83ea10",
    "assignerShortName": "VulnCheck",
    "dateUpdated": "2026-09-26T13:23:20.418Z",
    "dateReserved": "2026-09-26T02:33:07.898Z",
    "datePublished": "2026-09-26T13:23:20.418Z",
    "state": "PUBLISHED"
  },
  "containers": {
    "cna": {
      "providerMetadata": {
        "orgId": "83251b91-4cc7-4094-a5c7-464a1b83ea10",
        "shortName": "VulnCheck",
        "dateUpdated": "2026-09-26T13:23:20.418Z"
      },
      "datePublic": "2026-09-12T00:00:00.000Z",
      "title": "vLLM before 0.29.0 Resource Exhaustion via Unbounded Media Materialization",
      "descriptions": [
        {
          "lang": "en",
          "value": "vLLM through 0.29.0 fetches and fully materializes remote or inline media before enforcing its documented media controls (the VLLM_MAX_AUDIO_CLIP_FILESIZE_MB compressed-audio size cap, default 25 MB, and the per-modality --limit-mm-per-prompt item limits). Across four ingress paths — the shared media-acquisition layer (HTTPConnection.get_bytes()/async_get_bytes()), the chat completions audio_url/base64 path, the batch speech runner, and the Rust frontend POST /tokenize route — the server reads the entire HTTP response body, base64-decodes the inline payload, or spawns one fetch/decode task per media part, and only then applies the limit (or, on some paths, never applies it). A remote attacker can therefore cause the API server or batch-runner process to allocate memory and consume outbound bandwidth proportional to an attacker-chosen body size or media item count before the request is rejected, resulting in pre-inference memory and bandwidth exhaustion (denial of service). The chat and batch surfaces require an API key when one is configured; the Rust frontend /tokenize route is unauthenticated by design. There is no code execution or data disclosure impact."
        }
      ],
      "affected": [
        {
          "vendor": "vllm-project",
          "product": "vllm",
          "defaultStatus": "unaffected",
          "versions": [
            {
              "version": "0",
              "status": "affected",
              "versionType": "semver",
              "lessThan": "0.29.0"
            },
            {
              "version": "0.29.0",
              "status": "unaffected",
              "versionType": "semver"
            }
          ]
        }
      ],
      "problemTypes": [
        {
          "descriptions": [
            {
              "lang": "en",
              "description": "Uncontrolled Resource Consumption",
              "cweId": "CWE-400",
              "type": "CWE"
            }
          ]
        }
      ],
      "references": [
        {
          "url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-p6g9-7v3x-m8mv",
          "name": "GitHub Security Advisory (GHSA-p6g9-7v3x-m8mv)",
          "tags": [
            "vendor-advisory"
          ]
        },
        {
          "url": "https://github.com/vllm-project/vllm/commit/752a3a504485790a2e8491cacbb35c137339ad34",
          "name": "Patch Commit",
          "tags": [
            "patch"
          ]
        },
        {
          "url": "https://www.vulncheck.com/advisories/vllm-before-0.29.0-resource-exhaustion-via-unbounded-media-materialization",
          "name": "VulnCheck Advisory: vLLM before 0.29.0 Resource Exhaustion via Unbounded Media Materialization",
          "tags": [
            "third-party-advisory"
          ]
        }
      ],
      "metrics": [
        {
          "format": "CVSS"
        },
        {
          "format": "CVSS",
          "cvssV3_1": {
            "version": "3.1",
            "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H",
            "attackVector": "NETWORK",
            "attackComplexity": "LOW",
            "privilegesRequired": "LOW",
            "userInteraction": "NONE",
            "scope": "UNCHANGED",
            "confidentialityImpact": "NONE",
            "integrityImpact": "NONE",
            "availabilityImpact": "HIGH",
            "baseScore": 6.5,
            "baseSeverity": "MEDIUM"
          }
        }
      ],
      "credits": [
        {
          "lang": "en",
          "value": "KernelClint",
          "type": "finder"
        },
        {
          "lang": "en",
          "value": "jperezdealgaba",
          "type": "coordinator"
        }
      ]
    }
  }
}