Vulnerability · Published Sep 16, 2026 · Assigned by GitHub_M
CVE-2026-57173vLLM: Unauthenticated audio decompression-bomb DoS in /v1/chat/completions
Medium 6.5
In plain English
A medium-severity resource allocation without limits in vllm from vllm-project. There are no confirmed reports of exploitation so far. The chance it is exploited in the next 30 days is estimated at 0.7% (higher than 54% of all known vulnerabilities). Fixed versions are available - upgrade the affected package.
Official description
vLLM is an inference and serving engine for large language models. Prior to 0.24.0, the input_audio handling path for /v1/chat/completions calls AudioMediaIO.load_bytes or AudioMediaIO.load_file without passing VLLM_MAX_AUDIO_DECODE_DURATION_S to the shared audio decoder. An unauthenticated client can therefore submit a small compressed audio input that expands into a very large float32 PCM allocation, bypassing the duration guard already used by /v1/audio/transcriptions and causing an out-of-memory worker crash. Inline data URLs reach this path without being bounded by VLLM_AUDIO_FETCH_TIMEOUT. The issue affects deployments serving an audio-capable model, and authentication changes only the deployment-specific reachability. This issue is fixed in version 0.24.0.
Affected packages & fixed versions
| Ecosystem | Package | Fixed in | Advisory |
|---|---|---|---|
| PyPI | vllm | 0.24.0 | GHSA-hcwq-8wjf-3gcr |
| PyPI | vllm | 0.24.0 | PYSEC-2026-4179 |
Affected products (as reported)
| Vendor | Product | Versions |
|---|---|---|
| vllm-project | vllm | < 0.24.0 |
Weakness: CWE-770 · Vector: CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H
References
- https://github.com/vllm-project/vllm/security/advisories/GHSA-hcwq-8wjf-3gcr x_refsource_CONFIRM
- https://github.com/vllm-project/vllm/pull/45908 x_refsource_MISC
- https://github.com/vllm-project/vllm/commit/3d20275bb4d434f53055c3c0b645fd8bb072965e x_refsource_MISC
- https://github.com/vllm-project/vllm/releases/tag/v0.24.0 x_refsource_MISC