Service interruption via unbounded completion prompt fan-out
Published Aug 13, 2026 · Updated Aug 13, 2026
Resource exhaustion in vLLM 0.19.0 through 0.25.1 allows remote authenticated users to interrupt service via crafted prompt lists. The /v1/completions request model accepts an unbounded outer prompt list, which preprocessing expands before serving allocates one engine generator and response slot per element. An authenticated API credential and access to that endpoint are required; one request can starve shared CPU, memory, scheduler, engine, and buffering capacity.
Summary
What happened
Resource exhaustion in vLLM 0.19.0 through 0.25.1 allows remote authenticated users to interrupt service via crafted prompt lists. The /v1/completions request model accepts an unbounded outer prompt list, which preprocessing expands before serving allocates one engine generator and response slot per element. An authenticated API credential and access to that endpoint are required; one request can starve shared CPU, memory, scheduler, engine, and buffering capacity.
The record
- CVE
- CVE-2026-73559
- Published
- Aug 13, 2026
- Updated
- Aug 13, 2026
- Vendor
- vLLM Project
- Product
- vLLM
- Classifications
- CWE-400, T1499.004
- Attack vector
- network
- Privileges
- authenticated
Timeline
How it unfolded
- Aug 13, 2026CVE publishedPublication date reported by the CVE source.
- Aug 13, 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=>= 0.19.0, < 0.26.0
What conditions does exploitation require?
What is affected?
Published CVSS scores
CVSS describes severity. EPSS estimates exploitation probability.
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
- Oversized completion prompt-list proof of vulnerabilityproof of concept · demonstrated
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