---
title: "CVE-2026-61539: Xinference vulnerable to remote code execution via unsafe `eval()` in Llama3 tool-call parsing"
url: https://erroneo.us/security/cve/CVE-2026-61539/
description: "CVE-2026-61539 (xinference (PyPI)): Xinference is an inference API for running open-source, speech, and multimodal models. In 2.5.0 and earlier, Xinference…"
type: vulnerability-record
section: "Security Digest"
date_published: 2026-08-21
date_modified: 2026-10-08
publisher: erroneo.us
author: erroneo.us Security Desk
cve: "CVE-2026-61539"
kev: false
cvss: 10
epss: 0.01226
license: All rights reserved. Quote with attribution and a link.
---

# CVE-2026-61539: Xinference vulnerable to remote code execution via unsafe `eval()` in Llama3 tool-call parsing

**Signal:** Critical and remotely reachable. No known exploitation yet, but it is network-reachable without credentials. (Computed from CISA KEV, EPSS and CVSS; not an editorial verdict.)

## Should you act?

- No confirmed exploitation. It is not in CISA’s KEV catalog and CISA’s SSVC assessment records no known exploitation. That can change quickly after a public write-up; re-check if you are affected.
- EPSS 1.2% , percentile 69: FIRST’s model estimates a 1.2% chance of exploitation activity in the next 30 days, higher than 69% of all scored CVEs (moderate). EPSS is a probability, not a verdict.
- Remotely reachable with no login and no user interaction. If the vulnerable service is exposed to the internet, prioritize it.
- Only applies if your dependency tree includes xinference (PyPI) <= 2.5.0, fixed in 2.7.0 . Check your lockfile, including transitive dependencies.

## What it is

Xinference is an inference API for running open-source, speech, and multimodal models. In 2.5.0 and earlier, Xinference passes attacker-influenced Llama3 tool-call output to eval() in xinference/model/llm/toolparsers/llama3toolparser.py and xinference/model/llm/utils.py. Requests to /v1/chat/completions with a tools field flow through xinference/api/restfulapi.py, xinference/model/llm/transformers/core.py, handlechatresultnonstreaming(), and postprocesscompletion() before extracttoolcalls() or evalllama3chatarguments() evaluates the model-generated Python expression. An unauthenticated remote attacker can influence that output through a crafted prompt and execute commands in the Xinference server process context. This issue is fixed in version 2.7.0.

## Affected packages

- PyPI `xinference` <= 2.5.0 → fixed: 2.7.0

## Affected products (CVE record)

- xorbitsai inference: < 2.7.0

## Severity

- GitHub_M CVSS 3.1: 10 `CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H`
- GitHub CVSS 3.1: 10.0 `CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H`

## Exploitation

- CISA KEV: not listed (no confirmed in-the-wild exploitation known to CISA).
- CISA SSVC assessment: exploitation none , automatable yes , technical impact total .
- FIRST EPSS (model of 2026-10-21): 1.23% probability of exploitation activity in the next 30 days; percentile 69.

## Sources

- [GitHub Advisory GHSA-x2rj-828p-hx9m](https://github.com/advisories/GHSA-x2rj-828p-hx9m)
- [CVE.org record CVE-2026-61539](https://www.cve.org/CVERecord?id=CVE-2026-61539)
- [NVD entry CVE-2026-61539](https://nvd.nist.gov/vuln/detail/CVE-2026-61539)
- [github.com](https://github.com/xorbitsai/inference/security/advisories/GHSA-x2rj-828p-hx9m)
- [github.com](https://github.com/xorbitsai/inference/pull/4786)
- [github.com](https://github.com/xorbitsai/inference/commit/1b3d220f342ce68d34cec4586d9409d457dadc42)
- [github.com](https://github.com/xorbitsai/inference/releases/tag/v2.7.0)

---
Source page: https://erroneo.us/security/cve/CVE-2026-61539/ · Security Digest: https://erroneo.us/security/ · More: https://erroneo.us/llms.txt