# CVE-2026-115915: vLLM image format validation bypass in media processing

Source: https://junglewise.ai/threats/cve-2026-115915-vllm-is-an-inference-and-serving-engine-for-large-language  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-10-26. Updated: 2026-10-26  
License: CC BY 4.0 with a link back

CVE-2026-115915 · Severity: high · CVSS 8.1 · Published 2026-10-26

vLLM is an inference engine for large language models that processes images and other media as part of multimodal requests. A flaw in image handling allows attackers to upload specially crafted image files (such as PostScript/EPS payloads) that can trigger command execution on the server if Ghostscript is installed, or cause denial of service through resource exhaustion when certain versions of the Pillow library are present.

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-2026-115915 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-115915 being exploited in the wild.

**How severe is CVE-2026-115915?** High severity, with a CVSS score of 8.1, published 26 October 2026.

**Which products and versions are affected?** vllm (PyPI) < 0.32.0, ECOSYSTEM:introduced=0,fixed=0.32.0, ECOSYSTEM:introduced=0.30.0,fixed=0.32.0; vLLM Project vllm prior to 0.32.0.

**Is there a fix for CVE-2026-115915?** Yes. vllm (PyPI): upgrade to 0.32.0 or later. Until then: Restrict multimodal image upload paths via network-level controls or authentication policies until upgrading; disable or firewall external renderer dependencies such as Ghostscript if not required for image processing workflows. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-115915 attacked?** It is attacked over the network, in a software library.

**Am I affected by CVE-2026-115915?** 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 0.32.0 or later

  ```sh
  pip install "vllm>=0.32.0"
  ```


Restrict multimodal image upload paths via network-level controls or authentication policies until upgrading; disable or firewall external renderer dependencies such as Ghostscript if not required for image processing workflows.

## Technical details

vLLM's ImageMediaIO passes untrusted image bytes to Pillow's Image.open() without restricting decoder selection via a format allowlist, allowing PostScript/EPS payloads to be dispatched to external renderers. The vulnerability affects image loading paths including parse_image() and fetch_image(), and also impacts the Kimi fused-vision component's _to_pil() function. Exploitation requires network access to submit image-capable multimodal requests; impact chains through CVE-2024-29510 (Ghostscript RCE) or CVE-2026-59203 (Pillow DoS) depending on installed dependencies.

## Affected products

- vLLM Project vLLM prior to 0.32.0

## Identifiers

- CVE-2026-115915
- GHSA-h3rc-j82p-2c3h
- PYSEC-2026-4346

## Timeline

- 2026-10-26: disclosed
- 2026-10-05: patched

## References

- [https://github.com/vllm-project/vllm/commit/710ac56e691d4713feca50a43e6c440f07d373fd](https://github.com/vllm-project/vllm/commit/710ac56e691d4713feca50a43e6c440f07d373fd)
- [https://github.com/vllm-project/vllm/pull/60022](https://github.com/vllm-project/vllm/pull/60022)

## 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 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 Phi-4 multimodal audio encoder GPU memory exhaustion](https://junglewise.ai/threats/cve-2026-115913-vllm-is-an-inference-and-serving-engine-for-large-language)