CVE-2026-63632
Received Received - Intake

Out-of-Bounds Read in ONNX Version Converter

Publication date: 2026-08-18

Last updated on: 2026-08-18

Assigner: GitHub, Inc.

Description
Open Neural Network Exchange (ONNX) is an open standard for machine learning interoperability. From 1.3.0 until 1.22.0, onnx.version_converter.convert_version() can perform an out-of-bounds read in Gemm_7_6::adapt_gemm_7_6() in onnx/version_converter/adapters/gemm_7_6.h when a Gemm node has input tensors with fewer than two dimensions because B_shape[1], A_shape[0], or A_shape[1] is accessed without a rank check, potentially causing a process crash during an opset 7 to 6 downgrade. This issue is fixed in version 1.22.0.

CVSS Scores

EPSS Scores

Probability:
Percentile:

Meta Information

Published
2026-08-18
Last Modified
2026-08-18
Generated
2026-08-18
AI Q&A
2026-08-18
EPSS Evaluated
N/A
NVD
CVE-2026-63632
EUVD
EUVD-2026-60944

Affected Vendors & Products

Showing 2 associated CPEs
Vendor Product Version / Range
onnx onnx 1.3.0
onnx onnx From 1.3.0 (inc) to 1.22.0 (inc)

Helpful Resources

Exploitability

CWE
CWE Icon
KEV
KEV Icon
CWE ID Description
CWE-125 The product reads data past the end, or before the beginning, of the intended buffer.

Attack-Flow Graph

AI Quick Actions

Instant insights powered by AI
Executive Summary

This vulnerability is an out-of-bounds read in the ONNX library's version converter. It occurs when converting models from opset 7 to 6 using onnx.version_converter.convert_version(). The issue arises when a Gemm node has input tensors with fewer than two dimensions, causing the code to access array indices without checking tensor ranks first. This can lead to a process crash during conversion.

Detection Guidance

This vulnerability affects ONNX versions 1.3.0 to 1.22.0 during opset downgrades. To detect it, check the installed ONNX version using pip show onnx or python -c "import onnx; print(onnx.__version__)". If the version is between 1.3.0 and 1.21.0, the system is vulnerable.

Impact Analysis

If you use ONNX models and perform version downgrades from opset 7 to 6, this vulnerability could cause your application to crash unexpectedly. This might disrupt machine learning workflows, data processing pipelines, or any system relying on ONNX model conversions.

Mitigation Strategies

Upgrade ONNX to version 1.22.0 or later using pip install --upgrade onnx. Avoid downgrading opsets to version 7 if using affected versions. If downgrades are necessary, validate input tensor dimensions before conversion.

Chat Assistant

Ask questions about this CVE
Hi! I’m here to help you understand CVE-2026-63632. Ask me anything about the vulnerability, its impact, or mitigation strategies.
0/70

EPSS Chart