What is the CVSS score of CVE-2026-24146?

CVE-2026-24146 has a CVSS 3.1 base score of 7.5 (High). CVSS vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H. EPSS exploitation probability: 0.0%.

Which versions of Nvidia are affected by CVE-2026-24146?

NVIDIA Triton Inference Server contains a remote denial-of-service vulnerability (CVE-2026-24146) that allows unauthenticated attackers to crash inference services through malformed requests,…

How to fix or mitigate CVE-2026-24146?

Within 24 hours: Inventory all Triton Inference Server deployments and document current versions (particularly those pre-r26.02); isolate Triton instances from untrusted networks where possible. Within 7 days: Implement request validation at the load balancer/API gateway layer to restrict input sizes and output counts; deploy network segmentation to limit Triton exposure; establish monitoring for abnormal request patterns and service crashes. Within 30 days: Monitor NVIDIA security advisories for r26.02 release or patch availability; prepare and test deployment procedures for version upgrades; evaluate alternative inference platforms as contingency if patch timeline extends.

Nvidia CVE-2026-24146

EUVD-2026-19749 HIGH

Memory Allocation with Excessive Size Value (CWE-789)

2026-04-07 nvidia

Nvidia Denial Of Service

7.5

CVSS 3.1 · NVD

Severity by source

NVD PRIMARY

7.5 HIGH

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

Primary rating from NVD · only source for this CVE.

CVSS VectorNVD

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

Attack Vector

Network

Attack Complexity

Low

Privileges Required

None

User Interaction

None

Scope

Unchanged

Confidentiality

None

Integrity

None

Availability

High

Lifecycle Timeline

Re-analysis Queued

Apr 16, 2026 - 17:22 vuln.today

cvss_changed

EUVD ID Assigned

Apr 07, 2026 - 18:00 euvd

EUVD-2026-19749

Analysis Generated

Apr 07, 2026 - 18:00 vuln.today

CVE Published

Apr 07, 2026 - 17:11 nvd

HIGH 7.5

DescriptionCVE.org

NVIDIA Triton Inference Server contains a vulnerability where insufficient input validation and a large number of outputs could cause a server crash. A successful exploit of this vulnerability might lead to denial of service.

AnalysisAI

NVIDIA Triton Inference Server crashes when processing inference requests with insufficient input validation combined with large output counts, enabling remote denial of service without authentication (CVSS 7.5, EPSS data not available). The vulnerability affects all versions prior to r26.02, with no public exploit identified at time of analysis. Unauthenticated remote attackers can exploit this flaw with low complexity (AV:N/AC:L/PR:N) to completely disrupt machine learning inference services.

Technical ContextAI

NVIDIA Triton Inference Server is an open-source inference serving platform that provides GPU-accelerated deployment of AI models across frameworks including TensorFlow, PyTorch, and ONNX. This vulnerability is classified as CWE-789 (Memory Allocation with Excessive Size Value), indicating the server fails to properly validate input parameters before allocating resources for inference outputs. When an attacker crafts requests specifying excessively large output dimensions, the insufficient bounds checking allows memory allocation attempts that exceed safe thresholds, triggering a server crash. The affected component is cpe:2.3:a:nvidia:triton_inference_server, which processes inference requests over network protocols (HTTP/gRPC). The lack of input sanitization on output size parameters represents a fundamental input validation failure in the request handling pipeline, where untrusted user-controlled data directly influences resource allocation decisions without proper constraint enforcement.

RemediationAI

Upgrade NVIDIA Triton Inference Server to version r26.02 or later, which contains input validation improvements that properly constrain output size parameters and prevent excessive memory allocation attempts. Download the patched release from NVIDIA NGC (nvcr.io/nvidia/tritonserver:26.02-py3) or build from the corresponding GitHub release tag if using a custom deployment. For environments unable to immediately upgrade, implement network-level rate limiting and request size restrictions on Triton's HTTP (default port 8000) and gRPC (default port 8001) endpoints to mitigate exploitation attempts, though this provides only partial protection against crafted low-volume attacks. Review and restrict network access to inference endpoints using firewall rules or service mesh policies, limiting exposure to trusted clients only. Monitor server logs for abnormal termination events and unusual output dimension values in inference requests. Complete remediation guidance and version-specific patches are documented in NVIDIA's security bulletin at https://nvidia.custhelp.com/app/answers/detail/a_id/5816.