CVE-2026-53923: vLLM is an inference and serving engine for large language models (LLMs). From 0.5.5 until 0.23.1rc0…
- Occurred
- Jun 22, 2026
- Published
- Jun 22, 2026
- Source
- NVD CVE
Summary
vLLM is an inference and serving engine for large language models (LLMs). From 0.5.5 until 0.23.1rc0, integer truncation of tensor dimensions in vLLM's GGUF dequantize kernels (csrc/quantization/gguf/gguf_kernel.cu) causes partial tensor processing. The output tensor is allocated at full size via torch::empty (uninitialized memory), but the dequantize CUDA kernel processes only a truncated number of elements. The unfilled portion of the output tensor retains whatever was previously in GPU memory. In multi-tenant inference deployments, this residual GPU memory may contain tensor data from other users' inference requests, constituting information disclosure. This vulnerability is fixed in 0.23.1rc0.
Tags
- CVE
- CVE-2026-53923
- cvss:HIGH
Classification
Rationale: vLLM vulnerability allows information disclosure of other users' data via residual GPU memory in multi-tenant deployments.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8