---
id: CVE-2026-55093
title: >-
  Tract is a tiny, no-nonsense, self-contained TensorFlow and ONNX inference
  toolkit
summary: >-
  Tract is a tiny, no-nonsense, self-contained TensorFlow and ONNX inference
  toolkit. Prior to 0.21.16, 0.22.2, and 0.23.1, tract-nnef uses unchecked usize
  multiplication in nnef/src/tensors.rs read_tensor for attacker-controlled
  tensor di…
severity: medium
cvss: 6.1
cvssVector: 'CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:L/I:N/A:H'
cwe:
  - CWE-125
  - CWE-190
vendor: sonos
product: tract
affected:
  - tract < 0.21.16
  - 'tract >= 0.22.0, < 0.22.2'
  - 'tract >= 0.23.0, < 0.23.1'
patched:
  - tract-nnef 0.23.1
  - tract-nnef 0.22.2
  - tract-nnef 0.21.16
published: '2026-09-14'
updated: '2026-09-14'
sourceUpdated: '2026-09-14T20:16:47.290'
source: NVD
sourceUrl: 'https://nvd.nist.gov/vuln/detail/CVE-2026-55093'
references:
  - url: >-
      https://github.com/sonos/tract/commit/34c7df2c9bd2a36583e09b52f3e6319bf23102e8
    label: security-advisories@github.com
  - url: >-
      https://github.com/sonos/tract/commit/828ec20f74c14c0923c7d7f0cff062ed222d630a
    label: security-advisories@github.com
  - url: >-
      https://github.com/sonos/tract/commit/ee5f019d39fb97f7d4431b883c3ce97b74c28a64
    label: security-advisories@github.com
  - url: 'https://github.com/sonos/tract/releases/tag/0.21.16'
    label: security-advisories@github.com
  - url: 'https://github.com/sonos/tract/releases/tag/0.22.2'
    label: security-advisories@github.com
  - url: 'https://github.com/sonos/tract/releases/tag/v0.23.1'
    label: security-advisories@github.com
  - url: 'https://github.com/sonos/tract/security/advisories/GHSA-x5mv-8wgw-29hg'
    label: security-advisories@github.com
  - url: 'https://github.com/sonos/tract/security/advisories/GHSA-x5mv-8wgw-29hg'
    label: 134c704f-9b21-4f2e-91b3-4a467353bcc0
  - url: 'https://github.com/advisories/GHSA-x5mv-8wgw-29hg'
tags:
  - nvd
  - cve.org
  - exploit-available
  - ghsa
  - rust
exploitAvailable: true
ssvc:
  exploitation: poc
  automatable: 'no'
  technicalImpact: partial
  timestamp: '2026-09-14T20:09:04.689487Z'
ecosystem: rust
ingestedAt: '2026-06-19T03:39:00.798Z'
epss: 0.00181
epssPercentile: 0.06846
---

## Overview

Tract is a tiny, no-nonsense, self-contained TensorFlow and ONNX inference toolkit. Prior to 0.21.16, 0.22.2, and 0.23.1, tract-nnef uses unchecked usize multiplication in nnef/src/tensors.rs read_tensor for attacker-controlled tensor dimensions, the allocation size, and the reported tensor length. Loading a crafted NNEF archive through model_for_path or model_for_read reaches the default DatLoader and can make the wrapped size check accept a small allocation while data/src/tensor.rs as_slice_unchecked creates a much larger logical slice. Model construction through as_uniform can then read beyond the heap allocation and disclose adjacent data, and later access can terminate the process with a segmentation fault. The affected dense numeric tensor path does not include the independently guarded bool, String, or block-quant paths, and no out-of-bounds write or code execution was demonstrated. This issue is fixed in versions 0.21.16, 0.22.2, and 0.23.1.

## Remediation

Refer to the linked advisories for vendor-supplied fixes and affected version ranges.

## Package advisory (CVE-2026-55093)

Affected packages:

- `tract-nnef >= 0.23.0, < 0.23.1`
- `tract-nnef >= 0.22.0, < 0.22.2`
- `tract-nnef < 0.21.16`

Patched in:

- `tract-nnef 0.23.1`
- `tract-nnef 0.22.2`
- `tract-nnef 0.21.16`

Source: https://github.com/advisories/GHSA-x5mv-8wgw-29hg