# CVE-2026-90555: vLLM audio transcription endpoint denial of service via forged FLAC headers

Source: https://junglewise.ai/threats/cve-2026-90555-vllm-audio-transcription-endpoint-denial-of-service-via-forged  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-09-12. Updated: 2026-09-26  
License: CC BY 4.0 with a link back

CVE-2026-90555 · Severity: medium · CVSS 6.5 · Published 2026-09-12

vLLM is an open-source API server for running large language and speech models. When configured with audio transcription capabilities, it fails to properly validate audio sample rate headers, allowing authenticated users to crash the entire API server with a small malicious audio file. This causes a denial of service affecting all customers relying on that vLLM instance.

Technologies: [Vllm](https://junglewise.ai/threats/technologies/vllm). Vendors: [Vllm](https://junglewise.ai/threats/vendors/vllm).

## Quick answers

**Is CVE-2026-90555 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-90555 being exploited in the wild.

**How severe is CVE-2026-90555?** Medium severity, with a CVSS score of 6.5, published 12 September 2026.

**Which products and versions are affected?** vLLM Project vLLM before 0.28.0.

**Is there a fix for CVE-2026-90555?** A patch was published on 2026-09-12: Fixed in vLLM 0.28.0 Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-90555 attacked?** It is attacked over the network, in a hosted service.

**Am I affected by CVE-2026-90555?** Check whether you run Vllm. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

## Technical details

The vulnerability exists in vLLM's transcription endpoint (/v1/audio/transcriptions) audio decoding logic. The code validates audio duration by reading sample rate and frame count directly from untrusted FLAC container headers without verifying header integrity, then uses these values to compute a duration guard (frames / sample_rate). An authenticated attacker can submit a malicious FLAC file with forged headers specifying an extremely high sample rate (e.g., 655350 Hz) in a small file, causing the duration check to pass while the actual decode operation allocates gigabytes of memory based on the header values. The vulnerability only manifests on the primary soundfile/libsndfile code path; the PyAV fallback correctly counts decoded samples. Preconditions include: valid API credentials, a transcription-capable model deployed, the vllm[audio] extra installed, and no upstream proxy with body size limits. An authenticated attacker can repeatedly crash the shared API server process with requests as small as 3.2 MiB, causing service-wide denial of service across all API clients. A fix is available in version 0.28.0 and later.

## Affected products

- vLLM Project vLLM before 0.28.0

## Timeline

- 2026-08-28: disclosed: GitHub Security Advisory GHSA-99f2-hwrc-gvq8 published
- 2026-09-12: patched: Fixed in vLLM 0.28.0

## References

- [https://github.com/vllm-project/vllm/security/advisories/GHSA-99f2-hwrc-gvq8](https://github.com/vllm-project/vllm/security/advisories/GHSA-99f2-hwrc-gvq8)
- [https://www.vulncheck.com/advisories/vllm-before-0.28.0-denial-of-service-via-audio-header](https://www.vulncheck.com/advisories/vllm-before-0.28.0-denial-of-service-via-audio-header)

## Related threats

- [vLLM denial of service in DeepSeek-OCR image processing](https://junglewise.ai/threats/cve-2026-115914-vllm-is-an-inference-and-serving-engine-for-large-language)
- [vLLM heap overflow in audio transcription IAMF parser](https://junglewise.ai/threats/cve-2026-109311-vllm-before-0-29-0-passes-uploaded-audio-to-the-pyav-ffmpeg-av)
- [vLLM Qwen video backend memory exhaustion via unbounded sampling parameters](https://junglewise.ai/threats/cve-2026-105758-vllm-is-an-inference-and-serving-engine-for-large-language)
- [vLLM structured-output request handling denial of service](https://junglewise.ai/threats/cve-2026-105757-vllm-is-an-inference-and-serving-engine-for-large-language)
- [vLLM inference engine input validation bypass in scale-out](https://junglewise.ai/threats/cve-2026-105754-vllm-is-an-inference-and-serving-engine-for-large-language)