OpenAI’s Codex keyboard ships, and early users say it’s pricey but fun
APPSO · wechat · 2026-07-24
OpenAI’s Codex keyboard has finally shipped, and the first wave of user hands-on reports is in. The small device packs multiple mechanical keys, a joystick, a dial, and touch input, and early reactions are split between “expensive” and “surprisingly fun.”
What users like
- Build quality and key feel are widely praised, with a premium, Apple-like unboxing experience.
- The device makes it easier to switch between multiple agents than using a mouse and full keyboard.
- OpenAI’s voice updates and Codex workflows make a voice-plus-controller setup feel especially efficient.
What the device actually does
- Six translucent Agent Keys can be mapped to different Codex tasks and show task state with colors.
- Command Keys handle accept/reject/new chat/custom actions/voice input.
- The joystick can be assigned to four Skills, and the dial controls reasoning level by default.
Early caveats and hacks
- A few early users have already reported hardware issues, including a non-working joystick.
- Power users are remapping nearly everything, from approval flows to build actions and multi-agent threads.
- One user even repurposed the microphone key as a cross-app voice input shortcut using Wispr Flow, and built a plugin to control the keyboard’s lighting from Codex.
The broader takeaway: if you only run one chat, Codex Micro is mostly a fancy macro pad; if you juggle several agents at once, the physical state lights and dedicated controls may actually save real attention.
More from coding & agent
- Claude Code vs Codex: the harness may matter more than the model — al_kRicha · 2026-07-24
- Which model should power each agent task in production? — Logical-Silver-272 · 2026-07-24
- Offloop says its 4-person multi-agent harness beat Claude Code and Codex on GDPval — rohanpaul_ai · 2026-07-24
- Graph engineering, in one line: deterministic multi-agent loops become DAGs — vincent_koc · 2026-07-24
- BackSearch’s development was led by @ranko3000 with feedback from several collaborators — rosstaylor90 · 2026-07-24
- Default Codex CLI with GPT-5.5 scores 92.3% on XBOW, but the paper says the harness matters — evilsocket · 2026-07-24