> ## Documentation Index
> Fetch the complete documentation index at: https://docs.llmeasy.ru/llms.txt
> Use this file to discover all available pages before exploring further.

# Connect the VS Code Codex extension to LLMEasy with a custom model

> Configure a custom Base URL and API Key for the Codex extension in VS Code.

The VS Code Codex extension does not require a separate `Base URL` and API Key field inside VS Code. It reads your local Codex configuration, so configure the LLMEasy provider in Codex `config.toml`.

<Info>
  If you do not see `Base URL` or API Key fields in VS Code settings, that is expected. Configure model requests for the VS Code Codex extension through the local Codex configuration.
</Info>

## Prerequisites

* VS Code is installed
* The **Codex - OpenAI's coding agent** extension is installed
* You have a LLMEasy API Key (<a href={"https://www.llmeasy.ru/register"}>sign up and get one</a>)
* You have copied an available model ID from the **GPT Key group** in the <a href={"https://www.llmeasy.ru/pricing"}>model plaza</a>

<Warning>
  Do not put your API Key in project files, `AGENTS.md`, sample `.env` files, or public repositories. The configuration below should stay in your local Codex configuration directory.
</Warning>

## Setup methods

<Tabs>
  <Tab title="Recommended: use the Setup dialog">
    After you create an API Key, the console opens a **Setup** dialog. From that dialog, copy **API Key** and **Base URL**, then choose a Codex setup method.

    <Steps>
      <Step title="Select Codex">
        Select **Codex** under **Available tools**.
      </Step>

      <Step title="Choose a setup method">
        You can choose:

        * **One-click Bash Configuration**: copy the command and run it in your local terminal. Use this for quick setup.
        * **CC Switch Integration**: import the provider into CC Switch. Use this when you manage multiple tools.
        * **Manual Configuration**: copy API Key and Base URL, then write them into `config.toml`.
      </Step>

      <Step title="Restart VS Code">
        After the configuration is written, restart VS Code or run **Developer: Reload Window** so the Codex extension reads the local configuration again.
      </Step>
    </Steps>
  </Tab>

  <Tab title="Manual configuration">
    <Steps>
      <Step title="Find config.toml">
        The Codex configuration file is usually located at:

        | System        | Configuration file                      |
        | ------------- | --------------------------------------- |
        | macOS / Linux | `~/.codex/config.toml`                  |
        | Windows       | `C:\Users\YOUR_USER\.codex\config.toml` |
        | Windows + WSL | `~/.codex/config.toml` inside WSL       |

        If the file does not exist, create it.
      </Step>

      <Step title="Add the LLMEasy provider">
        Add the configuration below to `config.toml`. If the file already contains the same top-level fields, merge the configuration instead of writing duplicate `model_provider` entries.

        ```toml theme={null}
        model_provider = "custom"
        model = "gpt-5.5"
        review_model = "gpt-5.4"
        model_reasoning_effort = "high"
        model_context_window = 1000000
        model_auto_compact_token_limit = 900000
        windows_wsl_setup_acknowledged = true

        [model_providers]
        [model_providers.custom]
        name = "LLM Easy"
        base_url = "https://www.llmeasy.ru/v1"
        wire_api = "responses"
        requires_openai_auth = true
        experimental_bearer_token = "YOUR_API_KEY"
        request_max_retries = 4
        stream_max_retries = 8
        stream_idle_timeout_ms = 300000
        supports_websockets = false
        ```

        Replace `YOUR_API_KEY` with your LLMEasy API Key.
      </Step>

      <Step title="Confirm the key fields">
        Check these fields first:

        | Field                       | Value                                        |
        | --------------------------- | -------------------------------------------- |
        | `base_url`                  | `https://www.llmeasy.ru/v1`                  |
        | `wire_api`                  | `responses`                                  |
        | `model`                     | A model ID copied from the **GPT Key group** |
        | `experimental_bearer_token` | Your LLMEasy API Key                         |

        <Warning>
          For the VS Code Codex extension, do not convert `auth.json` into an API Key-only structure. `auth.json` keeps the official Codex login state; the third-party API Key belongs in `experimental_bearer_token`.
        </Warning>
      </Step>

      <Step title="Restart VS Code">
        After editing the file, close the current Codex session and restart VS Code.

        You can also run this from the command palette:

        ```text theme={null}
        Developer: Reload Window
        ```
      </Step>
    </Steps>
  </Tab>
</Tabs>

## Why keep the official login state?

The VS Code Codex extension depends on Codex app-side capabilities such as the extension UI, session management, and VS Code integration. Keep the official login state while routing model requests through LLMEasy.

In this configuration:

| Setting                                      | What it does                                                                 |
| -------------------------------------------- | ---------------------------------------------------------------------------- |
| `requires_openai_auth = true`                | Keeps the official Codex login state                                         |
| `experimental_bearer_token = "YOUR_API_KEY"` | Uses your LLMEasy API Key for model requests                                 |
| `base_url = "https://www.llmeasy.ru/v1"`     | Sends model requests to LLMEasy's OpenAI-compatible / Responses API endpoint |

## Verify the setup

After setup, open the Codex extension in VS Code and send a short test message.

If you get a normal response, the configuration is active.

If the request fails, check:

1. `base_url` is `https://www.llmeasy.ru/v1`
2. `wire_api` is `responses`
3. `experimental_bearer_token` has been replaced with your real API Key
4. `model` comes from the **GPT Key group** in the <a href={"https://www.llmeasy.ru/pricing"}>model plaza</a>
5. Old `OPENAI_API_KEY` or `OPENAI_BASE_URL` environment variables are not overriding the configuration
6. VS Code has been restarted or reloaded

## Related docs

* [Codex setup guide](/en/ai-tools/codex)
* [How to keep the official Codex login state with a third-party API](/en/faq/codex/official-login-third-party-api)
* [How to configure Codex CLI config.toml](/en/faq/codex/config-toml)
* [What are model\_provider, base\_url, and wire\_api in Codex CLI?](/en/faq/codex/model-provider-base-url-wire-api)