Service interruption via unbounded audio upload read
Published Jul 6, 2026 · Updated Jul 6, 2026
Uncontrolled resource consumption in vLLM 0.22.0 through 0.23.0 allows attackers to interrupt service via oversized audio uploads. The transcription and translation routes call request.file.read() without a size bound, fully materializing the upload before _preprocess_speech_to_text() applies the configured compressed-file limit. Exploitation requires exposed speech-to-text routes, a compatible model, and any required API key; memory pressure can terminate the process under deployment limits or concurrent requests.
Summary
What happened
Uncontrolled resource consumption in vLLM 0.22.0 through 0.23.0 allows attackers to interrupt service via oversized audio uploads. The transcription and translation routes call request.file.read() without a size bound, fully materializing the upload before _preprocess_speech_to_text() applies the configured compressed-file limit. Exploitation requires exposed speech-to-text routes, a compatible model, and any required API key; memory pressure can terminate the process under deployment limits or concurrent requests.
The record
- CVE
- CVE-2026-55646
- Published
- Jul 6, 2026
- Updated
- Jul 6, 2026
- Vendor
- vLLM Project
- Product
- vLLM
- Classifications
- CWE-400, CWE-770, T1499.004
- Attack vector
- network
- Privileges
- authenticated
Timeline
How it unfolded
- Jul 6, 2026CVE publishedPublication date reported by the CVE source.
- Jul 6, 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=>= 0.22.0, < 0.24.0
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
- audio_upload_size_precheck_probe.pyproof 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