# CVE-2026-114053: vLLM remote code execution via mm_processor_kwargs.code_revision

Source: https://junglewise.ai/threats/cve-2026-114053-vllm-from-0-7-3-before-0-31-0-forwards-the-request-supplied-mm  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-10-21. Updated: 2026-10-21  
License: CC BY 4.0 with a link back

CVE-2026-114053 · Severity: high · CVSS 8.1 · Published 2026-10-21

vLLM is an open-source language model serving framework that powers AI API endpoints. When configured with remote code execution enabled, an attacker can send a specially crafted request to the unauthenticated /tokenize endpoint to execute arbitrary Python code on the vLLM server, gaining full control of the API process. This affects deployments that serve models with dynamic processors and have trust_remote_code enabled.

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-2026-114053 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-114053 being exploited in the wild.

**How severe is CVE-2026-114053?** High severity, with a CVSS score of 8.1, published 21 October 2026.

**Which products and versions are affected?** vllm (PyPI) >= 0.7.3, < 0.31.0, ECOSYSTEM:introduced=0.7.3,fixed=0.31.0; vLLM Project vllm 0.7.3 before 0.31.0.

**Is there a fix for CVE-2026-114053?** Yes. vllm (PyPI): upgrade to 0.31.0 or later. Until then: Disable trust_remote_code when serving models, or protect the /tokenize endpoint with authentication middleware and external network access controls as a defense-in-depth measure. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-114053 attacked?** It is attacked over the network, in a software library.

**Am I affected by CVE-2026-114053?** 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 0.31.0 or later

  ```sh
  pip install "vllm>=0.31.0"
  ```


Disable trust_remote_code when serving models, or protect the /tokenize endpoint with authentication middleware and external network access controls as a defense-in-depth measure.

## Technical details

The vulnerability exists in vLLM's multimodal request handling, where untrusted mm_processor_kwargs.code_revision parameters are forwarded directly to Transformers' AutoProcessor.from_pretrained without filtering. When trust_remote_code is enabled, an attacker controlling a model repository revision can inject an attacker-controlled processor module that executes at import time. The /tokenize endpoint lacks authentication middleware protection, making it accessible without API key credentials even when configured.

## Affected products

- vLLM Project vLLM 0.7.3 before 0.31.0

## Identifiers

- CVE-2026-114053
- GHSA-h3rc-6mm3-gc2m

## Timeline

- 2026-10-06: disclosed
- 2026-10-21: patched

## References

- [https://github.com/vllm-project/vllm/commit/570735520942b6428ae98f1eb442c6d88adae55e](https://github.com/vllm-project/vllm/commit/570735520942b6428ae98f1eb442c6d88adae55e)
- [https://github.com/vllm-project/vllm/security/advisories/GHSA-h3rc-6mm3-gc2m](https://github.com/vllm-project/vllm/security/advisories/GHSA-h3rc-6mm3-gc2m)

## 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)