vLLM through 0.29.0 fetches and fully materializes remote or inline media before enforcing its documented media controls (the VLLM_MAX_AUDIO_CLIP_FILESIZE_MB compressed-audio size cap, default 25 MB, and the per-modality --limit-mm-per-prompt item limits). Across four ingress paths — the shared media-acquisition layer (HTTPConnection.get_bytes()/async_get_bytes()), the chat completions audio_url/base64 path, the batch speech runner, and the Rust frontend POST /tokenize route — the server reads the entire HTTP response body, base64-decodes the inline payload, or spawns one fetch/decode task per media part, and only then applies the limit (or, on some paths, never applies it). A remote attacker can therefore cause the API server or batch-runner process to allocate memory and consume outbound bandwidth proportional to an attacker-chosen body size or media item count before the request is rejected, resulting in pre-inference memory and bandwidth exhaustion (denial of service). The chat and batch surfaces require an API key when one is configured; the Rust frontend /tokenize route is unauthenticated by design. There is no code execution or data disclosure impact.
Metrics
Affected Vendors & Products
Advisories
No advisories yet.
Fixes
Solution
No solution given by the vendor.
Workaround
No workaround given by the vendor.
References
History
Sat, 26 Sep 2026 13:30:00 +0000
| Type | Values Removed | Values Added |
|---|---|---|
| Description | vLLM through 0.29.0 fetches and fully materializes remote or inline media before enforcing its documented media controls (the VLLM_MAX_AUDIO_CLIP_FILESIZE_MB compressed-audio size cap, default 25 MB, and the per-modality --limit-mm-per-prompt item limits). Across four ingress paths — the shared media-acquisition layer (HTTPConnection.get_bytes()/async_get_bytes()), the chat completions audio_url/base64 path, the batch speech runner, and the Rust frontend POST /tokenize route — the server reads the entire HTTP response body, base64-decodes the inline payload, or spawns one fetch/decode task per media part, and only then applies the limit (or, on some paths, never applies it). A remote attacker can therefore cause the API server or batch-runner process to allocate memory and consume outbound bandwidth proportional to an attacker-chosen body size or media item count before the request is rejected, resulting in pre-inference memory and bandwidth exhaustion (denial of service). The chat and batch surfaces require an API key when one is configured; the Rust frontend /tokenize route is unauthenticated by design. There is no code execution or data disclosure impact. | |
| Title | vLLM before 0.29.0 Resource Exhaustion via Unbounded Media Materialization | |
| First Time appeared |
Vllm
Vllm vllm |
|
| Weaknesses | CWE-400 | |
| CPEs | cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:* | |
| Vendors & Products |
Vllm
Vllm vllm |
|
| References |
| |
| Metrics |
cvssV3_1
|
Projects
Sign in to view the affected projects.
Status: PUBLISHED
Assigner: VulnCheck
Published:
Updated: 2026-09-26T13:23:20.418Z
Reserved: 2026-09-26T02:33:07.898Z
Link: CVE-2026-100650
No data.
Status : Received
Published: 2026-09-26T14:16:47.523
Modified: 2026-09-26T14:16:47.523
Link: CVE-2026-100650
No data.
OpenCVE Enrichment
Updated: 2026-09-26T17:15:14Z
Weaknesses