Unauthenticated server crash via empty prompt assertion
Published Sep 17, 2024 · Updated Nov 20, 2025
Reachable assertion in the vLLM completions API before 0.5.5 allows remote attackers to crash the API server with an empty prompt. The request produces no prompt token IDs, yet _get_num_new_tokens accepts it until an assertion requires a positive count, killing the asynchronous engine. An unauthenticated network client can interrupt all inference served by the affected process; the demonstrated request uses a model that preserves the empty prompt.
Summary
What happened
Reachable assertion in the vLLM completions API before 0.5.5 allows remote attackers to crash the API server with an empty prompt. The request produces no prompt token IDs, yet _get_num_new_tokens accepts it until an assertion requires a positive count, killing the asynchronous engine. An unauthenticated network client can interrupt all inference served by the affected process; the demonstrated request uses a model that preserves the empty prompt.
The record
- CVE
- CVE-2024-8768
- Published
- Sep 17, 2024
- Updated
- Nov 20, 2025
- Vendor
- vLLM Project
- Product
- vLLM
- Classifications
- CWE-617, T1499.004
- Attack vector
- network
- Privileges
- unauthenticated
Timeline
How it unfolded
- Sep 17, 2024CVE publishedPublication date reported by the CVE source.
- Nov 20, 2025Record 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 <0.5.5
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.
Public exploit references
- Upstream empty-prompt reproduction requestproof 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