Introducing the Codex Micro

See how the Codex Micro, a customizable keyboard built with Work Louder for Codex, helps you move through a build without leaving your flow. In this demo, GPT-5.6 Sol creates a word game while the Micro handles voice dictation, task switching, reasoning controls, permission requests, and follow-up prompts.

Get yours: https://openai.com/supply/

Chapters:

00:00 Introducing the Codex Micro

00:24 Voice dictation

00:45 Skills, Plan mode, and reasoning

01:01 Switching between tasks

01:16 Notifications and permissions

01:28 Queueing follow-up work

01:49 Game result and customization

Yes, the Codex Micro is designed to be highly customizable, meaning its utility extends beyond just coding tasks.

While the video demonstrates the device being used to manage Codex projects (0:00 – 1:53), it explicitly mentions that all controls are configurable (1:53 – 2:03). This allows you to map the buttons and the analog stick to various workflows and actions that fit your personal needs. For instance, the creator notes you can map the controls to common skills, pull requests, or even creative triggers like “triggering my pet” (2:01 – 2:03).

Essentially, because the interface is programmable, you can adapt the Codex Micro to streamline any software or digital process you use frequently, not just programming.

OpenAI's first hardware is Codex Micro, a keyboard for managing AI agents

The Codex Micro is OpenAI’s first-ever commercial hardware release, developed in partnership with custom keyboard maker Work Louder. Launched in July 2026, it is a $230 tactile macropad and control deck designed specifically for developers to navigate, orchestrate, and direct OpenAI’s Codex AI coding agents without relying heavily on a traditional keyboard or mouse. [1, 2, 3, 4, 5]

Key Specifications & Layout

Form Factor: A compact, premium mechanical macropad with an illuminated translucent shell.
Keys: 13 fully customizable mechanical shortcut keys equipped with dynamic RGB backlighting.
Control Sticks: Features a physical analog joystick and a customizable rotary dial.
Compatibility: Native integration out-of-the-box for both macOS and Windows platforms. [2, 6, 7, 8]

Core Functionality

Reasoning Adjustment: The rotary dial physically adjusts Codex’s AI reasoning effort on the fly, balancing swift tasks with deep computational thinking. [6, 9]
Thread Management: The physical joystick allows developers to smoothly toggle between multiple open AI agent workflows and pinned chat threads. [6, 7]
Action Shortcuts: Buttons are mapped to instant agent commands like creating pull requests, accepting/rejecting diffs, branching threads, or running terminal tests. [6, 10, 11, 12]
Voice Integration: Features a dedicated push-to-talk microphone button that streams voice prompts directly to the AI agent. [2, 10]
Status Monitoring: Smart RGB lighting acts as a visual status matrix, displaying if the agent is actively thinking, waiting for human approval, or idle. [8, 10]
Deep Customization: The companion layer mapping software allows users to configure up to 6 custom macro layers without external software vetting. [13]

If you are a developer looking to integrate this into your stack, I can provide more details on how to set up custom macro layers or outline the software requirements for the OpenAI Codex desktop app. Which would you prefer to explore first?