Replicas vs Codex cloud
Codex cloud is OpenAI's hosted home for Codex: included in paid ChatGPT plans, GitHub only for now, and you open the pull request from the finished task. Replicas runs Codex in an isolated VM per task on a ChatGPT plan, an OpenAI API key or Microsoft Foundry, opens the pull request itself, supports GitLab, and runs Claude Code and other agents beside it.
Checked against both vendors' docs and pricing as of November 9, 2026. This site is operated by Replicas.
Pick Replicas if
- You want Codex on an OpenAI API key or Microsoft Foundry; Codex cloud requires a ChatGPT sign-in.
- Your code is on GitLab, which Codex cloud doesn't support yet.
- You want work started from Slack, Linear, schedules or webhooks, with the agent opening the pull request.
- You want to send the same task to Codex and Claude Code and compare the results.
Pick Codex if
- Your team uses only Codex, works in GitHub and already has ChatGPT plans: Codex cloud is included.
- You mainly use Codex in the CLI or IDE and need cloud tasks occasionally.
Feature by feature
| Criterion | Replicas | Codex |
|---|---|---|
| Agents | ||
| Multi-harness support | YesClaude Code, Codex, Cursor, OpenCode, Pi and Muse Code; pick the agent per task | NoCodex only |
| Environment | ||
| Isolated cloud VM per task | YesEach workspace is its own isolated Linux VM (2, 4 or 8 vCPU) | YesCodex cloud runs tasks in OpenAI VMs (2 vCPU/8 GiB on Plus, 4 vCPU/16 GiB on Pro and Business) |
| Workflow | ||
| Pull request output | YesThe agent opens the pull request (GitHub) or merge request (GitLab) | PartlyYou open the PR from the finished task |
| Local IDE workflow | NoAgents run only in the cloud; the desktop app and CLI connect to cloud workspaces and can open one in VS Code | YesCLI, IDE extension and ChatGPT desktop app |
| Where work starts | ||
| Start from Slack | YesMentions, DMs and threads | CheckConfirm current Slack support with OpenAI |
| Start from Linear | YesAssign or mention on an issue | CheckConfirm with OpenAI |
| Start from GitHub | YesMention on issues and PRs; can act on failed checks and reviews | YesGitHub repositories; @codex review on pull requests |
| Start from GitLab | YesIssues and merge requests; one workspace can span GitHub and GitLab | NoGitHub only for now; GitLab is on OpenAI's roadmap |
| Models and billing | ||
| Bring your own model keys | YesAnthropic, OpenAI, AWS Bedrock, Microsoft Foundry, OpenRouter, Cursor and OpenAI-compatible keys | PartlyCodex cloud requires a ChatGPT sign-in; API keys work only in the CLI |
| Use your Claude or ChatGPT plan | YesClaude and ChatGPT subscriptions | YesChatGPT Plus, Pro, Business, Enterprise and Edu |
| Usage-based compute pricing | YesBilled per workspace minute by VM size; sleeping and archived workspaces cost nothing; seats are free | NoSeat-based ChatGPT plans from $20/mo |
| Nothing new to buy | NoA separate platform: Developer $50/mo, Team from $200/mo, each including that much usage | YesIncluded in paid ChatGPT plans |
| Security and governance | ||
| Self-hosted or dedicated deployment | YesSelf-hosted or dedicated single-tenant deployment on Enterprise | PartlyCodex cloud is OpenAI-hosted; the Apache-2.0 CLI runs anywhere |
| Scale | ||
| Parallel runs | YesRun many workspaces at once and compare them side by side | YesSeveral cloud tasks at once |
| Scheduled and event-driven runs | YesSchedules and GitHub, GitLab, Slack, Sentry and webhook events | PartlyThrough codex-action in GitHub workflows |
Sources: Replicas facts, Replicas pricing, Replicas enterprise, Replicas docs, Codex cloud, Codex pricing, Codex authentication.
Still deciding?
The comparison above is the same for everyone. The chooser weights it by your team size, the agents you use, where work starts, how you pay for models and what your security review needs, and tells you which of the two, or something else, fits.
Take the chooserCommon questions
Can Codex cloud use an API key?
OpenAI's docs say Codex cloud requires a ChatGPT sign-in, so an API key works only for the local CLI. Replicas runs Codex with a ChatGPT subscription, an OpenAI API key, Microsoft Foundry or an OpenAI-compatible endpoint.
Get a second opinion on your shortlist
Tell us what you're weighing and an engineer from the Replicas team, who operate this site, will reply by email. We'll point you to whichever option fits, including when that isn't Replicas.