OpenAI is stepping up its open source support. The company is expanding its [Codex](https://chatgpt.com/codex) for Open Source program with a concrete offer for maintainers: 6 months of [ChatGPT](https://chat.openai.com) Pro with Codex, conditional access to Codex Security, and API credits to automate maintenance workflows. All backed by a $1 million fund launched last year.
What the program includes
The program breaks down into three tracks, each targeting a different maintainer need:
- 6 months of ChatGPT Pro with Codex — For day-to-day coding, issue triage, code review, and maintenance workflows. This is OpenAI's flagship development tool, powered by GPT-5.4.
- Conditional access to Codex Security — For repositories that need deeper security coverage. Access is evaluated case by case by the OpenAI team, given GPT-5.4's capabilities.
- API credits via the Codex Open Source Fund — For projects that integrate Codex into PR review workflows, maintenance automation, releases, or other critical OSS tasks.
A $1 million fund backing it up
The Codex Open Source Fund, backed by $1 million, has been running for a year. It has already supported projects needing API credits, including teams using Codex to power GitHub pull request workflows. This new program track expands support beyond API credits: it gives access to OpenAI's productivity and security tools.
It's a strategic move. By offering premium tools to maintainers, OpenAI creates an ecosystem of developers familiar with its stack — ChatGPT, Codex, Codex Security — while genuinely contributing to the health of the global open source infrastructure.
Tool-agnostic: an open approach
Notable detail: OpenAI explicitly states that developers should code with whatever tools they prefer. The program supports maintainers whether they use Codex, OpenCode, Cline, pi, OpenClaw, or anything else.
This is a smart stance. Rather than forcing its ecosystem, OpenAI acknowledges ground-level reality: open source maintainers use a diverse range of tools — GitHub Copilot, Cursor, [Claude Code](https://docs.anthropic.com/en/docs/claude-code), and many others. By not conditioning aid on exclusive Codex usage, OpenAI gains credibility with the community.
Codex Security: case-by-case access
Access to Codex Security — the application security agent launched this week — isn't automatic. OpenAI evaluates each request individually. The stated reason: GPT-5.4's capabilities require special care for security workflows. In practice, OpenAI wants to ensure the tool is deployed responsibly.
For critical open source projects (cryptography libraries, major web frameworks, infrastructure tools), this is a significant opportunity. Codex Security already proved its value by discovering 14 CVEs in projects like OpenSSH, GnuTLS, and Chromium during its beta. The vLLM project is already using it in its regular workflow.
Who can apply?
OpenAI primarily targets core maintainers with write access and leaders of widely used public projects. But the program remains open: if your project doesn't perfectly fit the criteria but plays an important role in the ecosystem, OpenAI encourages you to apply and explain why.
Applications are submitted via the official form on the OpenAI website.
OpenAI vs Anthropic: the battle for open source
This program comes amid intense competition. Anthropic launched Claude Code Security on February 20 with a striking claim: over 500 zero-day vulnerabilities discovered in production open source projects. The impact was immediate — CrowdStrike, Okta, and Cloudflare stocks dropped the same day.
With Codex for Open Source, OpenAI responds on different ground: rather than just scanning code, it equips maintainers with comprehensive productivity tools. ChatGPT Pro for daily work, Codex for code, Codex Security for vulnerabilities — it's an integrated package aimed at making maintainers more effective across the board.
| OpenAI (Codex for OSS) | Anthropic (Claude Code Security) | |
|---|---|---|
| Maintainer offering | 6 months ChatGPT Pro + Codex + API credits | Claude Code Security access |
| Security | Codex Security (conditional access) | Claude Code Security (500+ zero-days) |
| Tool approach | Agnostic (use whatever you want) | Claude Code-centered |
| Underlying model | GPT-5.4 | Claude Opus 4.6 |
| CVEs discovered | 14 (beta) | 500+ (launch) |
OpenAI vs Anthropic open source programs comparison
What this means for the ecosystem
Open source maintainers do critical, often invisible work to keep the world's software infrastructure running. Most do it voluntarily, in their spare time, with limited tools. That two of the biggest AI companies are competing to equip them for free with premium tools is a strong signal.
Of course, the motivation isn't purely altruistic. Every maintainer who adopts ChatGPT Pro or Claude Code becomes a potential ambassador for the ecosystem. But the net result is positive: more security, more productivity, and concrete recognition of the essential role maintainers play.
If you maintain an open source project, now is the time to apply. Both programs are open and you have nothing to lose.
Tools mentioned in this article
Sources and references
Official sources:
- OpenAI - Codex for Open Source - openai.com
- Application form - openai.com
- Program terms - developers.openai.com
Check out our detailed reviews:
Follow AI news
Get the latest updates on AI tools and open source support programs.
No spam. Unsubscribe in 1 click.
