OpenAI just launched Codex in the ChatGPT desktop app for Linux in preview, and it hit 461 points with 315 comments on Hacker News. This isn't just another product launch — it's a signal that the AI coding assistant wars are expanding to the platform that developers actually use.
What's Available
The Linux preview supports:
- Ubuntu 24.04 LTS and 26.04 LTS
- Debian 13
- Fedora 43 and 44
- Both x64 and ARM64 architectures
-
.deband.rpmpackages
The desktop app combines ChatGPT, Work, and Codex in a single native experience, designed as a workspace for managing projects, working with files, running browser workflows, and using Codex alongside ChatGPT.
Why Linux Matters
Linux has long been the neglected stepchild of desktop AI tools. While Windows and macOS got native ChatGPT apps months ago, Linux users were stuck with browser tabs or CLI wrappers. For a platform that powers most of the world's servers, most AI infrastructure, and a disproportionate share of developer machines, this was always odd.
The Hacker News reaction tells the story:
"Finally. I've been running Codex through SSH tunnel hacks for months."
"ARM64 support out of the gate is genuinely impressive — that covers Raspberry Pi and Apple Silicon devs."
The Competitive Landscape
This launch comes at a critical moment in the AI coding tool race:
| Tool | Linux Support | Model | Key Feature |
|---|---|---|---|
| Codex (OpenAI) | ✅ Preview | GPT-5.6 | Desktop integration, browser workflows |
| Cursor | ✅ Native | Multi-model | IDE-native, codebase awareness |
| Claude Code | ✅ CLI | Claude Opus 5 | Terminal-first, agentic |
| Gemini Code Assist | ✅ VS Code | Gemini 3.7 | Free tier, large context |
| GitHub Copilot | ✅ VS Code | GPT-5.6 | Deep GitHub integration |
The pattern is clear: every major AI company now realizes that Linux developers are the canary in the coal mine. If your tool doesn't work on Linux, the most vocal developers won't use it, won't advocate for it, and won't recommend it to their companies.
What Codex on Linux Actually Does
Based on the announcement and community discussion:
- Project management: Native workspace for organizing coding projects
- File operations: Read, write, and modify files directly (not just suggest snippets)
- Browser workflows: Automate web-based tasks from within the desktop app
- Codex integration: Run Codex alongside ChatGPT for multi-turn coding sessions
- Architecture support: Both x64 and ARM64, covering everything from desktop Linux to Raspberry Pi
The ARM64 support is particularly interesting. It means you could theoretically run Codex on a Raspberry Pi 5 (with 8GB RAM) or an ARM-based cloud instance — though performance would obviously depend on whether the model runs locally or in the cloud.
The Community Response
The Hacker News thread reveals several themes:
Enthusiasm:
- Developers who've been waiting for native Linux support for months
- ARM users excited about being first-class citizens
- Debian 13 users noting it just launched and already has support
Skepticism:
- Concerns about telemetry and data collection
- Questions about whether this is Electron or native
- Comparisons to existing CLI tools (Claude Code, aider)
- Worry about vendor lock-in
Practical questions:
- How much RAM is needed for comfortable use?
- Does it work on Wayland?
- Can it access the local file system without sandboxing?
- Is there an API or is it GUI-only?
What This Means for the Broader Ecosystem
The Codex Linux launch is part of a larger trend: AI coding tools are becoming platform-native, not browser-based. This matters because:
- Latency-sensitive operations (like file editing) are better native than through a browser
- System integration (terminal, file system, package managers) requires native access
- Developer workflows happen on the desktop, not in a browser tab
- Offline capability (even partial) requires local installation
We're likely to see more of this. Google will probably bring Gemini Code Assist to a native Linux app. Anthropic already has Claude Code as a CLI tool. The race is now about who provides the best integrated experience, not just the best model.
The Real Winner: Linux Desktop
For years, people have been declaring "year of the Linux desktop" as a meme. But quietly, the AI coding revolution has made Linux desktop more relevant than ever:
- Most AI/ML workloads run on Linux
- Most cloud infrastructure is Linux
- Most AI researchers use Linux workstations
- ARM-based Linux devices (Pi, Jetson) are the edge AI platform
When the tools developers use every day support Linux natively, it stops being an "alternative" platform and becomes a primary one.
The Codex Linux preview is available now at openai.com/codex. If you're on Debian, Ubuntu, or Fedora, give it a try — and share your experience with the community.
Top comments (0)