CVE-2026-100651 (GCVE-0-2026-100651)

Vulnerability from cvelistv5 – Published: 2026-09-26 13:23 – Updated: 2026-09-26 13:23
VLAI
Title
vllm before 0.29.0 Denial of Service via Decoder Prompt Length Bypass
Summary
vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate. When the request contains a 'features' (multimodal) payload, vllm/entrypoints/serve/disagg/serving.py builds a multimodal EngineInput directly from the caller-supplied token_ids, and GenerateRequest.token_ids (vllm/entrypoints/serve/disagg/protocol.py) is not checked against model_config.max_model_len. For multimodal processors that report skip_prompt_length_check=True (for example Nemotron Parse, Whisper, and FireRedLID), InputProcessor._validate_prompt_len() returns immediately for both encoder and decoder prompts, so an overlong prompt becomes an EngineCoreRequest and reaches the worker input-batch copy into a fixed max_model_len-wide NumPy row. A client able to reach the endpoint on an affected model configuration can therefore submit an overlong token_ids list to trigger a worker failure and denial of service. Fixed in 0.29.0.
Severity
7.1 (High)
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N
6.5 (Medium)
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H
CWE
  • CWE-400 - Uncontrolled Resource Consumption
Assigner
References
Impacted products
Vendor Product Version CPE status
vllm-project vllm Affected: 0 , < 0.29.0 (semver)
Unaffected: 0.29.0 (semver)
    cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*
Create a notification for this product.
Date Public
2026-09-12 00:00
Credits
rexpository jperezdealgaba
Show details on NVD website
To clipboard 

{
  "containers": {
    "cna": {
      "affected": [
        {
          "defaultStatus": "unaffected",
          "packageURL": "pkg:pypi/vllm",
          "product": "vllm",
          "vendor": "vllm-project",
          "versions": [
            {
              "lessThan": "0.29.0",
              "status": "affected",
              "version": "0",
              "versionType": "semver"
            },
            {
              "status": "unaffected",
              "version": "0.29.0",
              "versionType": "semver"
            }
          ]
        }
      ],
      "cpeApplicability": [
        {
          "nodes": [
            {
              "cpeMatch": [
                {
                  "criteria": "cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:*",
                  "versionEndExcluding": "0.29.0",
                  "vulnerable": true
                }
              ],
              "negate": false,
              "operator": "OR"
            }
          ]
        }
      ],
      "credits": [
        {
          "lang": "en",
          "type": "reporter",
          "value": "rexpository"
        },
        {
          "lang": "en",
          "type": "coordinator",
          "value": "jperezdealgaba"
        }
      ],
      "datePublic": "2026-09-12T00:00:00.000Z",
      "descriptions": [
        {
          "lang": "en",
          "value": "vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate. When the request contains a \u0027features\u0027 (multimodal) payload, vllm/entrypoints/serve/disagg/serving.py builds a multimodal EngineInput directly from the caller-supplied token_ids, and GenerateRequest.token_ids (vllm/entrypoints/serve/disagg/protocol.py) is not checked against model_config.max_model_len. For multimodal processors that report skip_prompt_length_check=True (for example Nemotron Parse, Whisper, and FireRedLID), InputProcessor._validate_prompt_len() returns immediately for both encoder and decoder prompts, so an overlong prompt becomes an EngineCoreRequest and reaches the worker input-batch copy into a fixed max_model_len-wide NumPy row. A client able to reach the endpoint on an affected model configuration can therefore submit an overlong token_ids list to trigger a worker failure and denial of service. Fixed in 0.29.0."
        }
      ],
      "metrics": [
        {
          "cvssV4_0": {
            "attackComplexity": "LOW",
            "attackRequirements": "NONE",
            "attackVector": "NETWORK",
            "baseScore": 7.1,
            "baseSeverity": "HIGH",
            "privilegesRequired": "LOW",
            "subAvailabilityImpact": "NONE",
            "subConfidentialityImpact": "NONE",
            "subIntegrityImpact": "NONE",
            "userInteraction": "NONE",
            "vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N",
            "version": "4.0",
            "vulnAvailabilityImpact": "HIGH",
            "vulnConfidentialityImpact": "NONE",
            "vulnIntegrityImpact": "NONE"
          },
          "format": "CVSS"
        },
        {
          "cvssV3_1": {
            "attackComplexity": "LOW",
            "attackVector": "NETWORK",
            "availabilityImpact": "HIGH",
            "baseScore": 6.5,
            "baseSeverity": "MEDIUM",
            "confidentialityImpact": "NONE",
            "integrityImpact": "NONE",
            "privilegesRequired": "LOW",
            "scope": "UNCHANGED",
            "userInteraction": "NONE",
            "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H",
            "version": "3.1"
          },
          "format": "CVSS"
        }
      ],
      "problemTypes": [
        {
          "descriptions": [
            {
              "cweId": "CWE-400",
              "description": "Uncontrolled Resource Consumption",
              "lang": "en",
              "type": "CWE"
            }
          ]
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2026-09-26T13:23:21.141Z",
        "orgId": "83251b91-4cc7-4094-a5c7-464a1b83ea10",
        "shortName": "VulnCheck"
      },
      "references": [
        {
          "name": "GitHub Security Advisory (GHSA-3mqx-f33v-vgp9)",
          "tags": [
            "vendor-advisory"
          ],
          "url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-3mqx-f33v-vgp9"
        },
        {
          "name": "VulnCheck Advisory: vllm before 0.29.0 Denial of Service via Decoder Prompt Length Bypass",
          "tags": [
            "third-party-advisory"
          ],
          "url": "https://www.vulncheck.com/advisories/vllm-before-0.29.0-denial-of-service-via-decoder-prompt-length-bypass"
        }
      ],
      "title": "vllm before 0.29.0 Denial of Service via Decoder Prompt Length Bypass",
      "x_generator": {
        "engine": "vulncheck-endgame"
      }
    }
  },
  "cveMetadata": {
    "assignerOrgId": "83251b91-4cc7-4094-a5c7-464a1b83ea10",
    "assignerShortName": "VulnCheck",
    "cveId": "CVE-2026-100651",
    "datePublished": "2026-09-26T13:23:21.141Z",
    "dateReserved": "2026-09-26T02:33:07.899Z",
    "dateUpdated": "2026-09-26T13:23:21.141Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2",
  "vulnerability-lookup:meta": {
    "epss": {
      "cve": "CVE-2026-100651",
      "date": "2026-09-27",
      "epss": "0.00307",
      "percentile": "0.21045"
    },
    "nvd": {
      "cve": {
        "affected": [
          {
            "affectedData": [
              {
                "defaultStatus": "unaffected",
                "packageURL": "pkg:pypi/vllm",
                "product": "vllm",
                "vendor": "vllm-project",
                "versions": [
                  {
                    "lessThan": "0.29.0",
                    "status": "affected",
                    "version": "0",
                    "versionType": "semver"
                  },
                  {
                    "status": "unaffected",
                    "version": "0.29.0",
                    "versionType": "semver"
                  }
                ]
              }
            ],
            "source": "disclosure@vulncheck.com"
          }
        ],
        "cveTags": [],
        "descriptions": [
          {
            "lang": "en",
            "value": "vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate. When the request contains a \u0027features\u0027 (multimodal) payload, vllm/entrypoints/serve/disagg/serving.py builds a multimodal EngineInput directly from the caller-supplied token_ids, and GenerateRequest.token_ids (vllm/entrypoints/serve/disagg/protocol.py) is not checked against model_config.max_model_len. For multimodal processors that report skip_prompt_length_check=True (for example Nemotron Parse, Whisper, and FireRedLID), InputProcessor._validate_prompt_len() returns immediately for both encoder and decoder prompts, so an overlong prompt becomes an EngineCoreRequest and reaches the worker input-batch copy into a fixed max_model_len-wide NumPy row. A client able to reach the endpoint on an affected model configuration can therefore submit an overlong token_ids list to trigger a worker failure and denial of service. Fixed in 0.29.0."
          }
        ],
        "id": "CVE-2026-100651",
        "lastModified": "2026-09-26T14:16:47.663",
        "metrics": {
          "cvssMetricV31": [
            {
              "cvssData": {
                "attackComplexity": "LOW",
                "attackVector": "NETWORK",
                "availabilityImpact": "HIGH",
                "baseScore": 6.5,
                "baseSeverity": "MEDIUM",
                "confidentialityImpact": "NONE",
                "integrityImpact": "NONE",
                "privilegesRequired": "LOW",
                "scope": "UNCHANGED",
                "userInteraction": "NONE",
                "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H",
                "version": "3.1"
              },
              "exploitabilityScore": 2.8,
              "impactScore": 3.6,
              "source": "disclosure@vulncheck.com",
              "type": "Primary"
            }
          ],
          "cvssMetricV40": [
            {
              "cvssData": {
                "Automatable": "NOT_DEFINED",
                "Recovery": "NOT_DEFINED",
                "Safety": "NOT_DEFINED",
                "attackComplexity": "LOW",
                "attackRequirements": "NONE",
                "attackVector": "NETWORK",
                "availabilityRequirement": "NOT_DEFINED",
                "baseScore": 7.1,
                "baseSeverity": "HIGH",
                "confidentialityRequirement": "NOT_DEFINED",
                "exploitMaturity": "NOT_DEFINED",
                "integrityRequirement": "NOT_DEFINED",
                "modifiedAttackComplexity": "NOT_DEFINED",
                "modifiedAttackRequirements": "NOT_DEFINED",
                "modifiedAttackVector": "NOT_DEFINED",
                "modifiedPrivilegesRequired": "NOT_DEFINED",
                "modifiedSubAvailabilityImpact": "NOT_DEFINED",
                "modifiedSubConfidentialityImpact": "NOT_DEFINED",
                "modifiedSubIntegrityImpact": "NOT_DEFINED",
                "modifiedUserInteraction": "NOT_DEFINED",
                "modifiedVulnAvailabilityImpact": "NOT_DEFINED",
                "modifiedVulnConfidentialityImpact": "NOT_DEFINED",
                "modifiedVulnIntegrityImpact": "NOT_DEFINED",
                "privilegesRequired": "LOW",
                "providerUrgency": "NOT_DEFINED",
                "subAvailabilityImpact": "NONE",
                "subConfidentialityImpact": "NONE",
                "subIntegrityImpact": "NONE",
                "userInteraction": "NONE",
                "valueDensity": "NOT_DEFINED",
                "vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X",
                "version": "4.0",
                "vulnAvailabilityImpact": "HIGH",
                "vulnConfidentialityImpact": "NONE",
                "vulnIntegrityImpact": "NONE",
                "vulnerabilityResponseEffort": "NOT_DEFINED"
              },
              "source": "disclosure@vulncheck.com",
              "type": "Secondary"
            }
          ]
        },
        "published": "2026-09-26T14:16:47.663",
        "references": [
          {
            "source": "disclosure@vulncheck.com",
            "url": "https://github.com/vllm-project/vllm/security/advisories/GHSA-3mqx-f33v-vgp9"
          },
          {
            "source": "disclosure@vulncheck.com",
            "url": "https://www.vulncheck.com/advisories/vllm-before-0.29.0-denial-of-service-via-decoder-prompt-length-bypass"
          }
        ],
        "sourceIdentifier": "disclosure@vulncheck.com",
        "vulnStatus": "Received",
        "weaknesses": [
          {
            "description": [
              {
                "lang": "en",
                "value": "CWE-400"
              }
            ],
            "source": "disclosure@vulncheck.com",
            "type": "Primary"
          }
        ]
      }
    },
    "redhat_vex": {
      "aggregate_severity": "Moderate",
      "current_release_date": "2026-09-27T20:44:14+00:00",
      "cve": "CVE-2026-100651",
      "id": "CVE-2026-100651",
      "initial_release_date": "2026-09-26T13:23:21.141000+00:00",
      "product_status:known_affected": "27",
      "source": "Red Hat CSAF VEX",
      "status": "final",
      "title": "vllm: vLLM: Denial of Service via overlong multimodal inference requests",
      "url": "https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-100651.json",
      "version": "3"
    }
  }
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.





Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

Sightings

Author Source Type Date Other

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or observed by the user.
  • Confirmed: The vulnerability has been validated from an analyst's perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
  • Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
  • Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
  • Not confirmed: The user expressed doubt about the validity of the vulnerability.
  • Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.



Related by attack behaviour

Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.