codex/prs/bolinfest/PR-1693.md

PR #1693: Update config.md to clarify the configuration for Azure OpenAI provider

• URL: https://github.com/openai/codex/pull/1693
• Author: chenxizhang
• Created: 2025-07-26 23:00:35 UTC
• Updated: 2025-08-07 03:05:07 UTC
• Changes: +20/-3, Files changed: 1, Commits: 2

Description

(No description.)

Full Diff

diff --git a/codex-rs/config.md b/codex-rs/config.md
index c45d81180d..51ece5c793 100644
--- a/codex-rs/config.md
+++ b/codex-rs/config.md
@@ -65,17 +65,34 @@ base_url = "https://api.mistral.ai/v1"
 env_key = "MISTRAL_API_KEY"

-Note that Azure requires api-version to be passed as a query parameter, so be sure to specify it as part of query_params when defining the Azure provider: + +Note that Azure requires api-version to be passed as a query parameter, so be sure to specify it as part of query_params when defining the Azure provider, please copy the endpoint from Azure AI Foundry and then extract the base_url and api_version from the endpoint carefully. + +If you want to use codex-mini or other newer models that support the responses API, use the following configuration:

[model_providers.azure]
name = "Azure"
# Make sure you set the appropriate subdomain for this URL.
-base_url = "https://YOUR_PROJECT_NAME.openai.azure.com/openai"
-env_key = "AZURE_OPENAI_API_KEY"  # Or "OPENAI_API_KEY", whichever you use.
+base_url = "https://YOUR_PROJECT_NAME.openai.azure.com/openai" 
+env_key = "AZURE_OPENAI_API_KEY"  
query_params = { api-version = "2025-04-01-preview" }
+wire_api: "responses"

+If you want to use gpt-4.1 or o4-mini, which only support the chat API, use the following configuration: + +toml +[model_providers.azure] +name = "Azure" +# Make sure you set the appropriate subdomain for this URL. +base_url = "https://YOUR_PROJECT_NAME.openai.azure.com/openai/deployments/YOUR_DEPLOYMENT_NAME" +env_key = "AZURE_OPENAI_API_KEY" +query_params = { api-version = "2025-04-01-preview" } +wire_api: "chat" + + + It is also possible to configure a provider to include extra HTTP headers with a request. These can be hardcoded values (http_headers) or values read from environment variables (env_http_headers):

Review Comments

codex-rs/config.md

• Created: 2025-07-27 03:13:00 UTC | Link: https://github.com/openai/codex/pull/1693#discussion_r2233620285

@@ -71,7 +71,7 @@ Note that Azure requires `api-version` to be passed as a query parameter, so be
 [model_providers.azure]
 name = "Azure"
 # Make sure you set the appropriate subdomain for this URL.
-base_url = "https://YOUR_PROJECT_NAME.openai.azure.com/openai"
+base_url = "https://YOUR_PROJECT_NAME.openai.azure.com/openai" # If you are using `chat` as the `wire_api` (e.g., when using models like `gpt-4o` or `o3-mini`), the `base_url` should be like **https://YOUR_PROJECT_NAME.openai.azure.com/openai/deployments/YOUR_DEPLOYMENT_NAME**. Please note that `codex-mini` has been onboarded into Azure AI Foundry, when using this model, set `wire_api` to `responses` and use a `base_url` like **https://YOUR_PROJECT_NAME.openai.azure.com/openai**. Be sure to copy the exact URL from Azure AI Foundry, as it may vary by region or deployment method.

It seems more appropriate to add wire_api = "chat" to this example and to put this comment there.

Please put the comment before wire_api and break it up at 80 or 100 cols like normal source code so it is easier to read when this is rendered as Markdown. The current implementation causes horizontal scrolling in the Markdown, which you can verify here in GitHub using the Display the rich diff button.