How many CVEs have patched _convert_image_mode?

1 published CVEs changed _convert_image_mode in vllm-project/vllm, in 1 distinct patches totalling 5 changed lines, between 2026 and 2026.

What is the latest CVE that changed _convert_image_mode?

CVE-2026-12491 (Vllm: vllm: image exif rotation & png trns transparency not normalized, causing mismatch between model input and expectations), published 2026-06-17, changed lines 68 to 75 of vllm/multimodal/media/image.py.

How does _convert_image_mode compare to the rest of vllm-project/vllm?

vllm-project/vllm has 83 CVEs with a located fix in total; 1 of them changed _convert_image_mode.

What do the line numbers refer to?

Line numbers are in the fixed version of the file, as the upstream patch left it. A CVE whose fix was backported to several branches is listed once, with the lines from the upstream commit.

RadicalNotion.AI

Log in Create an account

vllm-project/vllm

Function

_convert_image_mode

Name: _convert_image_mode in vllm-project/vllm: CVE fixes
Creator: RadicalNotion.AI
License: https://creativecommons.org/licenses/by-nc/4.0/

vllm/multimodal/media/image.py

Last patched June 17, 2026

1

CVEs

That changed this function

1

Distinct patches

Backports counted once

5

Lines changed

Added plus removed

2026

Years

By CVE publication date

Overview

_convert_image_mode in vllm-project/vllm was changed by 1 published CVE fixes in 2026, in 1 distinct patches that added or removed 5 lines in total.

vllm-project/vllm has 83 CVEs with a located fix in total, so 1% of them touched this function.

CVEs that patched it

Every published CVE whose fix changed _convert_image_mode, newest first, with the lines each fix touched.

Vllm: vllm: image exif rotation & png trns transparency not normalized, causing mismatch between model input and expectations
CVE-2026-12491 · lines 68–75 · +4 −1
Medium · CVSS 4.8
EPSS 0.2%2026-06-17

Get an alert when the next CVE fix lands in vllm-project/vllm, with the patched functions.

Monitor vllm CVEs

Other functions in these files

Functions in the same file that CVE fixes also changed.

Other hotspots in the repository

Functions in vllm-project/vllm changed by the most CVE fixes.

validate_xgrammar_grammar4 CVEs · vllm/v1/structured_output/backend_xgrammar.py load_bytes3 CVEs · vllm/multimodal/media/audio.py _resolve_rust_frontend_path3 CVEs · vllm/envs.py load_file3 CVEs · vllm/multimodal/media/audio.py __init__3 CVEs · vllm/model_executor/models/kimi_k25.py _try_resolve_transformers3 CVEs · vllm/model_executor/models/registry.py

All patched functions in vllm-project/vllm

Frequently asked questions

Common questions about CVE fixes to _convert_image_mode.

How many CVEs have patched _convert_image_mode?: 1 published CVEs changed _convert_image_mode in vllm-project/vllm, in 1 distinct patches totalling 5 changed lines, between 2026 and 2026.
What is the latest CVE that changed _convert_image_mode?: CVE-2026-12491 (Vllm: vllm: image exif rotation & png trns transparency not normalized, causing mismatch between model input and expectations), published 2026-06-17, changed lines 68 to 75 of vllm/multimodal/media/image.py.
How does _convert_image_mode compare to the rest of vllm-project/vllm?: vllm-project/vllm has 83 CVEs with a located fix in total; 1 of them changed _convert_image_mode.
What do the line numbers refer to?: Line numbers are in the fixed version of the file, as the upstream patch left it. A CVE whose fix was backported to several branches is listed once, with the lines from the upstream commit.

References

Fix locations are computed by RadicalNotion.AI from the upstream patch diff of each CVE. CVE data is sourced from the CVE Program.

Track vllm-project/vllm CVEs

Get an alert when a new CVE lands in this repository, with the patched functions and AI-written remediation guidance.

Create an account See pricing