.. meta::
    :description: Set up Flower Endeavor 1.0 with Codex CLI and the ChatGPT/Codex Desktop app on macOS using a Flower API key and model catalog.
    :property=og:description: Set up Flower Endeavor 1.0 with Codex CLI and the ChatGPT/Codex Desktop app on macOS using a Flower API key and model catalog.

Use Endeavor with ChatGPT/Codex
===============================

.. note::

    You need a Flower API key with Endeavor access. Request access by filling out
    the `Endeavor 1.0 access form <https://flowerlabs.typeform.com/to/jlniHsuy>`_.

This guide connects Codex CLI and local Codex tasks in the ChatGPT/Codex
desktop app to ``flwrlabs/endeavor-1.0`` at ``https://api.flower.ai/v1``. The
commands below use macOS and its default shell, zsh. Before continuing, install
your preferred client using the `Codex CLI installation guide
<https://learn.chatgpt.com/docs/codex/cli>`_ or the `desktop app setup guide
<https://learn.chatgpt.com/docs/quickstart>`_.

For model details, see :doc:`endeavor`. To use OpenCode instead, see
:doc:`endeavor-opencode`.

1. Save the model catalog
-------------------------

Download :download:`flower-models.json <_static/flower-models.json>` to your
``Downloads`` folder, keeping that filename. This catalog makes **Endeavor**
available in the model selector and defines its capabilities.

.. note::

    The supplied catalog configures a 128,000-token context window and a
    96,000-token automatic compaction threshold for this client. These are
    client configuration values, not a universal limit on Endeavor API
    requests. See :ref:`Context window and limits <endeavor-context-limits>`
    for details.

In your terminal, copy the file to the Codex configuration directory:

.. code-block:: zsh

   mkdir -p "$HOME/.codex"
   cp "$HOME/Downloads/flower-models.json" "$HOME/.codex/flower-models.json"
   printf 'model_catalog_json = "%s/.codex/flower-models.json"\n' "$HOME"

Copy the printed ``model_catalog_json`` line for the next step.

2. Configure Codex
------------------

The CLI and desktop app share ``~/.codex/config.toml``. Create the file if
needed, back it up, and open it:

.. code-block:: zsh

   touch "$HOME/.codex/config.toml"
   cp -p "$HOME/.codex/config.toml" \
     "$HOME/.codex/config.toml.backup.$(date +%Y%m%d-%H%M%S)"
   open -a TextEdit "$HOME/.codex/config.toml"

Add or update the following settings, preserving unrelated configuration.
Keep the first four settings at the top level, before any ``[section]``.
Replace the ``model_catalog_json`` line with the one printed in
`1. Save the model catalog`_, then save the file. Update existing keys and sections rather than adding duplicates.

.. code-block:: toml

   model = "flwrlabs/endeavor-1.0"
   model_provider = "flower"
   model_reasoning_effort = "low"
   model_catalog_json = "/Users/YOUR_USERNAME/.codex/flower-models.json"

   [model_providers.flower]
   name = "Flower Labs"
   base_url = "https://api.flower.ai/v1"
   wire_api = "responses"
   env_key = "FLOWER_API_KEY"

For more options, see the `Codex configuration reference
<https://developers.openai.com/codex/config-reference/>`_.

3. Set your Flower API key
--------------------------

Follow :doc:`endeavor-api-key` to set ``FLOWER_API_KEY`` in your terminal.
For desktop use, also complete that guide's desktop setup before reopening
the app. Keep the terminal open for the next step.

4. Start using Endeavor
-----------------------

Codex CLI
~~~~~~~~~

In the same terminal, change to your project directory and start Codex:

.. code-block:: zsh

   codex

Endeavor is selected by default. Send a prompt such as
``Explain the structure of this project.`` Repeat the instructions in
:doc:`endeavor-api-key` when starting from a new terminal session.

ChatGPT/Codex desktop app on macOS
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

After completing the desktop key setup in :doc:`endeavor-api-key`, reopen
ChatGPT or Codex, start a new local Codex task, and select **Endeavor** in the
model selector. Send a prompt to begin.

Get help
--------

If you encounter any issues, feel free to post on
`Flower Discuss <https://discuss.flower.ai>`_ or
`Flower Slack <https://flower.ai/join-slack>`_. The Flower team will get back
to you soon.