Supported Model Aliases in the OpenAI Codex Plugin

The OpenAI Codex plugin supports exactly one model alias—spark—which internally maps to the full model identifier gpt-5.3-codex-spark.

The openai/codex-plugin-cc repository provides a command-line interface for interacting with Codex models. Understanding the supported model aliases helps you reference the correct model using shorthand syntax when invoking tasks. The plugin currently defines a single alias to simplify model selection across the CLI.

How the Model Alias System Works

The plugin implements a straightforward mapping system that translates user-friendly short names into the complete model identifiers required by the underlying API.

Alias Definition in the Source Code

The mapping is hardcoded in plugins/codex/scripts/codex-companion.mjs at line 72. Here, a Map object named MODEL_ALIASES stores the relationship between the alias and the concrete model:

const MODEL_ALIASES = new Map([["spark", "gpt-5.3-codex-spark"]]);

This single entry creates the association. When you specify --model spark, the CLI resolves this to gpt-5.3-codex-spark before sending the request to the Codex API.

Using the Spark Alias

You can invoke the model using either the short alias or the full identifier. Both approaches execute identical underlying logic.

Using the alias:

node scripts/codex-companion.mjs task --model spark --effort low "diagnose the failing test"

Using the full model name:

node scripts/codex-companion.mjs task --model gpt-5.3-codex-spark --effort low "diagnose the failing test"

The test suite in tests/commands.test.mjs specifically asserts that the spark alias correctly resolves to gpt-5.3-codex-spark, ensuring consistent behavior across plugin updates.

Implementation Details

The MODEL_ALIASES constant is referenced throughout the command-handling logic to validate user input. If you provide a value to --model that exists as a key in this Map, the plugin substitutes the corresponding value before constructing the API payload.

According to the source code analysis, no other aliases are defined in the current codebase. The README.md documents this behavior, recommending the spark alias as the primary shorthand for the gpt-5.3-codex-spark model.

Summary

• Single alias support: The Codex plugin defines only one model alias (spark).
• Concrete mapping: The alias maps to gpt-5.3-codex-spark as defined in plugins/codex/scripts/codex-companion.mjs.
• Flexible usage: You may use either the alias or the full model identifier interchangeably in CLI commands.
• Test coverage: The mapping is validated in tests/commands.test.mjs to prevent regression.

Frequently Asked Questions

What model aliases are supported by the Codex plugin?

The plugin supports exactly one alias: spark. This is the only shorthand defined in the MODEL_ALIASES Map within codex-companion.mjs.

How do I use the spark alias in a command?

Pass --model spark to any task command. For example: node scripts/codex-companion.mjs task --model spark --effort low "your prompt here". The plugin automatically expands this to the full model identifier.

Can I use the full model name instead of the alias?

Yes. You can specify --model gpt-5.3-codex-spark directly. Both the alias and the full name resolve to the same underlying model according to the mapping logic in the source code.

Is there a way to add custom model aliases?

The current implementation in openai/codex-plugin-cc does not support user-defined aliases. The MODEL_ALIASES Map is hardcoded with a single entry, and the CLI does not provide configuration options to extend this mapping.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}