Image input doesn't work for AoAI model

What version of Codex is running?

0.31.0

Which model were you using?

gpt-5,o4-mini

What platform is your computer?

Linux 5.15.146.1

What steps can reproduce the bug?

Config.toml

model_provider = "azure"
provider = "azure"
model = "gpt-5"
approvalMode="full-auto"
fullAutoErrorMode="ignore-and-continue"
approval_policy ="never"
sandbox_mode="danger-full-access"

[model_providers.azure]
name = "AzureOpenAI"
base_url = "https://xxxxxxx.cognitiveservices.azure.com/openai/deployments/gpt-5"
env_key = "AOAI_API_KEY"
query_params = { api-version = "2025-04-01-preview" }

Command

Codex -i text-image.png "summarize the image content"

What is the expected behavior?

Official doc example:

<img width="1565" height="1266" alt="Image" src="https://github.com/user-attachments/assets/08771c4f-bcdd-487b-8e3a-db779b68190e" />

What do you see instead?

The codex isn't aware of the input image. Is this related to aoai provider model?

<img width="1145" height="53" alt="Image" src="https://github.com/user-attachments/assets/42ea5240-a6c2-4efc-93b1-4eafb1596e69" />

<img width="1511" height="550" alt="Image" src="https://github.com/user-attachments/assets/8b88937d-219a-4d78-95f3-590b5ca05e28" />

<img width="1504" height="645" alt="Image" src="https://github.com/user-attachments/assets/1b5fe2ad-2fb1-404d-ac74-ac29a13e391e" />

Additional information

_No response_