# CVE-2026-105760: vLLM resource exhaustion in GLMGA video backend

Source: https://junglewise.ai/threats/cve-2026-105760-vllm-is-an-inference-and-serving-engine-for-large-language  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-10-05. Updated: 2026-10-05  
License: CC BY 4.0 with a link back

CVE-2026-105760 · Severity: medium · CVSS 5.3 · Published 2026-10-05

vLLM is an open-source inference engine for large language models that processes video input through a GLMGA backend. An attacker can submit requests with large video frame sampling parameters that cause excessive CPU and memory consumption in the shared media-loading executor, potentially disrupting service availability for other users.

Technologies: [vllm (PyPI)](https://junglewise.ai/threats/technologies/pypi-vllm), [vLLM Project vllm](https://junglewise.ai/threats/technologies/vllm-project-vllm). Vendors: [PyPI](https://junglewise.ai/threats/vendors/pypi), [vLLM Project](https://junglewise.ai/threats/vendors/vllm-project).

## Quick answers

**Is CVE-2026-105760 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-105760 being exploited in the wild.

**How severe is CVE-2026-105760?** Medium severity, with a CVSS score of 5.3, published 5 October 2026.

**Which products and versions are affected?** vllm (PyPI) >= 0.23.0rc2, < 0.30.0, ECOSYSTEM:introduced=0.23.0rc2,fixed=0.30.0; vLLM Project vllm prior to 0.30.0.

**Is there a fix for CVE-2026-105760?** Yes. vllm (PyPI): upgrade to 0.30.0 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-105760 attacked?** It is attacked over the network, in a software library.

**Am I affected by CVE-2026-105760?** Check whether you run vllm (PyPI) or vLLM Project vllm at an affected version. 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.

## How to fix

- vllm (PyPI): upgrade to 0.30.0 or later

  ```sh
  pip install "vllm>=0.30.0"
  ```


## Technical details

A request-level resource exhaustion vulnerability in vLLM's GLMGA video backend allows attackers to supply large fps and max_frames values via media_io_kwargs without strict resource limits. The GLMGA backend constructs a deduplicated frame-index list based on attacker input, enabling a small valid video to trigger disproportionate CPU and memory consumption in the shared media-loading executor. This is a network-accessible DoS vector in multi-tenant or shared inference scenarios.

## Affected products

- vLLM Project vLLM prior to 0.30.0

## Identifiers

- CVE-2026-105760
- GHSA-58v5-2m8f-94pr
- PYSEC-2026-4307

## Timeline

- 2026-09-04: patched: Security fix merged in PR #54935 to cap GLMGA video sampling
- 2026-10-05: disclosed

## References

- [https://github.com/vllm-project/vllm/commit/8b6de0eb9a09ef53f20cf06bd4d17ee264b9c2a7](https://github.com/vllm-project/vllm/commit/8b6de0eb9a09ef53f20cf06bd4d17ee264b9c2a7)
- [https://github.com/vllm-project/vllm/pull/54935](https://github.com/vllm-project/vllm/pull/54935)

## Related threats

- [vLLM chat template resource exhaustion DoS](https://junglewise.ai/threats/cve-2026-115918-vllm-is-an-inference-and-serving-engine-for-large-language)
- [vLLM: Permissive default CORS origins allow cross-origin requests to the vLLM HTTP server](https://junglewise.ai/threats/cve-2026-115917-vllm-is-an-inference-and-serving-engine-for-large-language)
- [vLLM off-by-one error in tokenizer max_token_id validation](https://junglewise.ai/threats/cve-2026-115916-vllm-is-an-inference-and-serving-engine-for-large-language)
- [vLLM image format validation bypass in media processing](https://junglewise.ai/threats/cve-2026-115915-vllm-is-an-inference-and-serving-engine-for-large-language)
- [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)