Codex CLI ignores configured model (gpt-5.4-mini) when starting a new session

What version of Codex CLI is running?

0.140.0

What subscription do you have?

ChatGPT Enterprise

Which model were you using?

gpt-5.4-mini

What platform is your computer?

Darwin 25.5.0 arm64 arm

What terminal emulator and version are you using (if applicable)?

Wezterm

Codex doctor report

{
  "schemaVersion": 1,
  "overallStatus": "ok",
  "codexVersion": "0.140.0",
  "checks": {
    "config.load": {
      "status": "ok",
      "summary": "config loaded",
      "details": {
        "CODEX_HOME": "/Users/<redacted-user>/.codex",
        "config.toml": "/Users/<redacted-user>/.codex/config.toml",
        "config.toml parse": "ok",
        "cwd": "/Users/<redacted-user>/src/github.com/<redacted-org>/<redacted-project>",
        "model": "gpt-5.4",
        "model provider": "openai",
        "sqlite home": "/Users/<redacted-user>/.codex"
      }
    },
    "auth.credentials": {
      "status": "ok",
      "summary": "auth is configured",
      "details": {
        "auth storage mode": "Auto",
        "stored API key": "false",
        "stored ChatGPT tokens": "true",
        "stored agent identity": "false",
        "stored auth mode": "chatgpt"
      }
    },
    "sandbox.helpers": {
      "status": "ok",
      "summary": "sandbox configuration is readable",
      "details": {
        "approval policy": "OnRequest",
        "filesystem sandbox": "restricted",
        "network sandbox": "enabled"
      }
    },
    "installation": {
      "status": "ok",
      "summary": "installation looks consistent",
      "details": {
        "managed by npm": "true"
      }
    },
    "runtime.provenance": {
      "status": "ok",
      "summary": "running npm on macos-aarch64",
      "details": {
        "platform": "macos-aarch64",
        "version": "0.140.0"
      }
    },
    "system.environment": {
      "status": "ok",
      "summary": "OS language ja-JP",
      "details": {
        "os": "Mac OS 26.5.1 [64-bit]",
        "os language": "ja-JP"
      }
    },
    "terminal.env": {
      "status": "ok",
      "summary": "terminal metadata was detected",
      "details": {
        "terminal": "WezTerm",
        "multiplexer": "zellij 0.44.3"
      }
    }
  }
}

What issue are you seeing?

1. Start Codex CLI.
2. Run /model.
3. Select gpt-5.4-mini.
4. Verify that ~/.codex/config.toml contains:

model = "gpt-5.4-mini"
model_reasoning_effort = "low"
model_verbosity = "low"

1. Exit Codex
2. Start a completely new session using: codex

the new session does not use the configured model.

The session starts with GPT-5.4 (medium) instead of the configured gpt-5.4-mini.
The configuration file remains unchanged and still contains gpt-5.4-mini.

What steps can reproduce the bug?

1. Start Codex CLI.
2. Run /model.
3. Select gpt-5.4-mini.
4. Verify that ~/.codex/config.toml contains:

model = "gpt-5.4-mini"
model_reasoning_effort = "low"
model_verbosity = "low"

1. Exit Codex.
2. Start a completely new session using:

codex

1. Run /status.

What is the expected behavior?

The newly created session should use:

Model: gpt-5.4-mini
Reasoning: low
Verbosity: low

Additional information

• gpt-5.4-mini is selectable and usable.
• The configuration is persisted correctly.
• The issue only occurs when creating a new session.
• This appears to be related to session initialization rather than model availability.
• If helpful, I can provide the exact /status output before and after restarting Codex.