CVE-2026-24220

A vulnerability in the NVIDIA TensorRT-LLM library has been found in the OpenAI-compatible inference API. An attacker can trigger a reachable assertion in the sampler thread, leading to denial of service.

NVIDIA TensorRT-LLM contains a vulnerability that could allow an attacker to trigger a write-what-where condition. Successful exploitation may lead to data tampering, denial of service, and information disclosure.

A vulnerability in NVIDIA TensorRT-LLM's inter-process communication layer allows an attacker with local same-user access to trigger deserialization. Successful exploitation could lead to code execution, information disclosure, data tampering, and denial of service.

A vulnerability in NVIDIA TensorRT-LLM during tensor deserialization allows an attacker to trigger a heap-based buffer overflow. This could lead to information disclosure, data tampering, or denial of service.

NVIDIA TensorRT-LLM contains a vulnerability in the gRPC server chat API endpoint, allowing a local attacker to trigger CWE-20 (improper input validation). Successful exploitation could lead to denial of service (DoS).

NVIDIA TensorRT-LLM contains a vulnerability in the OpenAI-compatible inference API, where an attacker could cause allocation of GPU resources without limits or throttling. A successful exploit of this vulnerability might lead to denial of service.

NVIDIA TensorRT-LLM for Linux contains a vulnerability that could allow an attacker to bypass authentication for a critical function. Successful exploitation may lead to code execution, data tampering, and information disclosure.

A vulnerability in NVIDIA TensorRT-LLM for Linux exists in the multimodal media fetching functions, allowing a network-accessible attacker to perform server-side request forgery (SSRF). Successful exploitation could lead to denial of service and information disclosure.

NVIDIA TensorRT-LLM for Linux contains a vulnerability in the restricted unpickler used for model weight deserialization, allowing a local, unauthenticated attacker to deserialize untrusted data. Successful exploitation could lead to code execution, privilege escalation, data tampering, and information disclosure.

A vulnerability in NVIDIA TensorRT-LLM for Linux within the disaggregated orchestrator component allows an attacker to read, write, or delete internal cluster state by sending requests to the FastAPI server. Successful exploitation could lead to information disclosure, data tampering, and denial of service.