Cross-tenant tensor disclosure via truncated CUDA counts
Published Jun 22, 2026 · Updated Jun 22, 2026
Information disclosure in vLLM 0.5.5 and later, before 0.23.1rc0, allows remote attackers to expose residual GPU tensor data via crafted GGUF models. The GGUF dequantization path passes 64-bit tensor element counts into 32-bit CUDA kernel parameters, so truncated launch bounds leave part of each torch::empty output tensor unwritten and populated with stale GPU memory. Exploitation requires a victim to load an attacker-crafted GGUF model with more than INT_MAX tensor elements in a multi-tenant inference deployment, where residual data from other users' requests can enter model computation.
Summary
What happened
Information disclosure in vLLM 0.5.5 and later, before 0.23.1rc0, allows remote attackers to expose residual GPU tensor data via crafted GGUF models. The GGUF dequantization path passes 64-bit tensor element counts into 32-bit CUDA kernel parameters, so truncated launch bounds leave part of each torch::empty output tensor unwritten and populated with stale GPU memory. Exploitation requires a victim to load an attacker-crafted GGUF model with more than INT_MAX tensor elements in a multi-tenant inference deployment, where residual data from other users' requests can enter model computation.
The record
- CVE
- CVE-2026-53923
- Published
- Jun 22, 2026
- Updated
- Jun 22, 2026
- Vendor
- vLLM Project
- Product
- vLLM
- Classifications
- CWE-200, CWE-681, T1204.002
- Attack vector
- network
- Privileges
- unauthenticated
Timeline
How it unfolded
- Jun 22, 2026CVE publishedPublication date reported by the CVE source.
- Jun 22, 2026Record updatedLatest update available in the CVE record.
Exploitability
Present is not the same as exploitable
Compare your product and version with the public record. A matching version still requires validation against your environment.
Is a vulnerable build present?
Compare these published version ranges with your installed build and any vendor patches.
- Affected versionversion=>= 0.5.5, < 0.23.1rc0
What conditions does exploitation require?
What is affected?
Attacks
What attackers are doing with it
Daily unique IPs observed by Shadowserver honeypots for known exploited vulnerabilities (KEVs). Missing observations do not establish an absence of attacks.
Weakness, pattern, technique
Public exploit references
No public exploit references are available in this record.
Labels summarize the accepted research assessment. They do not indicate a test against your environment.
Technologies
Your stack
See the directory against your own environment.
Your stack
Check the software in your environment
Book a demo to see how Hinoki identifies affected software and validates exploitability in your environment.
Book a demo