ollama - 💡(How to fix) Fix ollama launch codex-app writes legacy profile config rejected by Codex App [1 pull requests]

What is the issue?

ollama launch codex-app can make Codex App fail to start a chat with Codex 0.134+ / 0.135 because Ollama writes Codex's legacy profile config into ~/.codex/config.toml.

The generated config includes a root profile selector and profile table like:

profile = "ollama-launch-codex-app"

[profiles.ollama-launch-codex-app]
model = "..."
openai_base_url = "http://127.0.0.1:11434/v1/"
model_provider = "ollama-launch-codex-app"
model_catalog_json = "/Users/.../.codex/ollama-launch-models.json"

Codex App then reports a startup/chat error like:

Error starting chat
failed to resolve feature override precedence: legacy `profile = "ollama-launch-codex-app"` config is no longer supported; use `--profile ollama-launch-codex-app` with `ollama-launch-codex-app.config.toml` instead

This is related to, but distinct from, #16348. That issue's main report is for ollama launch codex / Codex CLI. This issue tracks the codex-app path specifically.

Steps to reproduce

1. Use Ollama 0.24.0+ with a recent Codex App / Codex runtime that rejects legacy profile config.

2. Run:

   ollama launch codex-app --model gemma4:31b

3. Open Codex App and start a chat.

4. Observe the legacy profile config error.

Expected behavior

ollama launch codex-app should configure Codex App without writing legacy profile selectors or [profiles.ollama-launch-codex-app] into ~/.codex/config.toml.

For Codex App, Ollama can keep the app selection in root model, model_provider, and model_catalog_json, and keep the provider in [model_providers.ollama-launch-codex-app].

Related PR

I opened #16364 for the Codex App side of this. #16351 appears to address the Codex CLI side.

OS

macOS

GPU

No response

CPU

No response

Ollama version

0.24.0