Is CVE-2026-115920 exploited in the wild?

No. Junglewise Threat Intelligence has no report of CVE-2026-115920 being exploited in the wild.

How severe is CVE-2026-115920?

Medium severity, with a CVSS score of 6.2, published 26 October 2026.

Which products and versions are affected?

ONNX 1.2.1 through 1.21.x; onnx (PyPI) >= 1.2.1, <= 1.21.0, ECOSYSTEM:introduced=1.2.1,fixed=1.22.0.

Is there a fix for CVE-2026-115920?

Yes. onnx (PyPI): upgrade to 1.22.0 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

How is CVE-2026-115920 attacked?

It is attacked over the network, in a software library.

Am I affected by CVE-2026-115920?

Check whether you run ONNX or onnx (PyPI) 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.

CVE-2026-115920: ONNX ConvTranspose shape inference denial of service

Q: Is there a fix for CVE-2026-115920?

Yes. onnx (PyPI): upgrade to 1.22.0 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

Q: How is CVE-2026-115920 attacked?

It is attacked over the network, in a software library.

Q: Am I affected by CVE-2026-115920?

Check whether you run ONNX or onnx (PyPI) 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.

CVE-2026-115920 · Severity: medium · CVSS 6.2 · Published 2026-10-26

ONNX is an open standard format for machine learning models. A flaw in shape inference for the ConvTranspose operator can cause the shape inference function to crash when processing a malformed model file, disrupting model loading and processing pipelines without exposing data or compromising system integrity.

Technologies: ONNX, onnx (PyPI). Vendors: ONNX, PyPI.

Quick answers

Is CVE-2026-115920 exploited in the wild?: No. Junglewise Threat Intelligence has no report of CVE-2026-115920 being exploited in the wild.
How severe is CVE-2026-115920?: Medium severity, with a CVSS score of 6.2, published 26 October 2026.
Which products and versions are affected?: ONNX 1.2.1 through 1.21.x; onnx (PyPI) >= 1.2.1, <= 1.21.0, ECOSYSTEM:introduced=1.2.1,fixed=1.22.0.
Is there a fix for CVE-2026-115920?: Yes. onnx (PyPI): upgrade to 1.22.0 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.
How is CVE-2026-115920 attacked?: It is attacked over the network, in a software library.
Am I affected by CVE-2026-115920?: Check whether you run ONNX or onnx (PyPI) 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

onnx (PyPI): upgrade to 1.22.0 or later
```
pip install "onnx>=1.22.0"
```

Technical details

The vulnerability exists in convTransposeShapeInference_opset11() which processes a weight tensor with fewer than three dimensions and no kernel_shape attribute, deriving an empty effective_kernel_shape and then indexing it with the input's spatial dimensions, causing an out-of-bounds access. An untrusted ONNX model that passes model validation can trigger a crash in onnx.shape_inference.infer_shapes(), leading to denial of service. The fix adds explicit validation to ensure input and weight tensors have minimum required dimensions before shape inference.

Affected products

ONNX ONNX 1.2.1 through 1.21.x

Identifiers

CVE-2026-115920
GHSA-6pwm-x34p-fvj2

Timeline

2026-10-26: disclosed
2026-05-19: patched: Fix committed to main branch

References

This page as Markdown. Free to reuse under CC BY 4.0 with a link back.